Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Research ArticleRobust FDI for a Class of Nonlinear NetworkedSystems with ROQs
An-quan Sun Lin Zhang Wen-feng Wang Jun Wang Tian-lin Niu and Jia-li Zhang
Air and Missile Defense College Air Force Engineering University Xirsquoan 710051 China
Correspondence should be addressed to Wen-feng Wang wwf981612163com
Received 9 September 2014 Accepted 3 November 2014 Published 20 November 2014
Academic Editor Zidong Wang
Copyright copy 2014 An-quan Sun et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
This paper considers the robust fault detection and isolation (FDI) problem for a class of nonlinear networked systems (NSs)with randomly occurring quantisations (ROQs) After vector augmentation Lyapunov function is introduced to ensure theasymptotically mean-square stability of fault detection system By transforming the quantisation effects into sector-boundedparameter uncertainties sufficient conditions ensuring the existence of fault detection filter are proposed which can reduce thedifference between output residuals and fault signals as small as possible under119867
infinframework Finally an example linearized from
a vehicle system is introduced to show the efficiency of the proposed fault detection filter
1 Introduction
With the wide use of computer network and the decreaseof networked techniquersquos cost structures of control systemsare changed The wires of traditional control systems werereplaced by computer network reserved for special or publicuse thus information can be transmitted via controllerssensors actuators and other system units [1] This type ofcontrol systems is used or will be used not only in large-scaledistributed systems (such as large-scale industrial controlsystems) but also in centralized small-scale local area systems(such as space vehicles airliners ships high-performancemotors etc) [2 3] For this type of control systems infor-mation like detection control coordinate instruction andso forth is transmitted via public network While functionslike estimation control diagnosis and so forth can be con-ducted among different network nodes distributively Controlsystems with at least one loop or several loops to make upclosed loop by networks are called networked systems (NSs)Compared to traditional control systems NSs have less wireswhich make them convenient to install and service In thissituation NSs can reduce systemsrsquo weight and size and theyare also convenient to extend As well as the advantagesbrought by NSs through sharing resources on network newchallenges and opportunities were brought to its systems
and control theories such as data transmission delay datadropout and immanent quantisation problemDuring recentdecades many famous journals about automatic control havepublished special issues on NSs [4ndash7] Specifically in 2003report on the development of automatic control made byAstrom and some famous scholars attracted unprecedentedattention to NSs also plenty of research results published[8] However compared to the comprehensive researches onNSsrsquo stability [9ndash11] control [12 13] and filtration [14ndash16]researches on NSsrsquo fault detection seem to be less
Fault detection is a technique to determine whether somecharacteristics or parameters of the system arouse largerdeviation which is unacceptable It is a key technique whichcan improve systemrsquos security and reliability and developedrapidly in the past thirty years Fault detection belongs tothe fault determination part in fault diagnosis procedure Atypical fault diagnosis procedure includes residual genera-tion residual evaluation separation strategy fault estimateand performance evaluation Firstly generate residual signalswhen fault occurs and the signals are sensitive to the fault androbust to unknown input and system parameters uncertainty[16 17] The most ideal condition to reach the above demandis to make residuals decoupled with unknown inputs relatedonly to faults But the decoupling conditions are usuallysevere and most systems are unable to reach In the latest
Hindawi Publishing CorporationMathematical Problems in EngineeringVolume 2014 Article ID 473901 8 pageshttpdxdoiorg1011552014473901
2 Mathematical Problems in Engineering
years a generally adopted approach is transforming the resid-ual generation problem into optimization problem Design afault detection filter (residual generator) which makes thetransfer functionrsquos 119867
infinbound norm from disturbance to
residual as small as possible but the transfer functionrsquos 119867infin
bound norm from (or other bound norms) fault to residualas large as possible Some existing researches demonstratethat there exists a unified solving method among differentoptimal indices [18 19] After the residuals are generatedselect a suitable residual evaluation function and a thresholdvalue Residual evaluation functions and threshold values canbe different in form but usually we take residual evaluationfunction over than threshold value as the judgment of faultalarm Here is the description of fault detection problem thisis what we care about in this paper After fault detection isdone faults are located with suitable separation strategy levelof the faults is estimated by kinds of approaches superiorityof the diagnosis algorithm is evaluated This is the completefault diagnosis procedure
Most of the existing fault detection theories are set upbased on traditional control systems whose data transmis-sions are without delay or with ideal data dropout amongsystem components The appearance of NSs brings newchallenges to traditional fault detection theories and out ofthis the main problem is that the signals transmitted vianetwork became imperfect including delay data dropoutquantisation effect and other signal changes not expectedduring the design Compared to traditional control systemsthe difficulty of NSsrsquo fault detection mainly depends on theimperfection of observation signals By now most of theexisting research results of NSsrsquo diagnosis detection resultsfocus on dealingwith network induced delay and few of themfocus on data dropout multipacket transmission or signalquantisation [20ndash23] In [24ndash27] the authors studied systemrsquosstability and control problem with information coder In[28] Fu and Xie discussed the logarithmic quantiser anddesigned the corresponding state feedback controller Yang etal in [29] designed a networked filter based on logarithmicquantiser In [30] Li et al studied NSsrsquo fault diagnosisproblem whose residual errors have quantisation errors andthen designed fault diagnosis filters respectively based onstatic and dynamic coders But so far to the best of theauthorsrsquo knowledge there are very few related results in theliterature that focus on quantisation problems for NSs withrandomly occurring quantisations (ROQs) This inspires ourcurrent research
In this paper the robust fault detection and isolation(FDI) problem is investigated for a class of nonlinearNSswithROQs A time-variant system with unknown disturbancesand nonlinear terms is studiedThe parameters of the systemare time-variant and satisfy norm-bounded condition andall the quantisers are assumed to be in the logarithmic typewith different quantisation lawsOn the other hand Bernoullidistributed stochastic sequence is used to decide which quan-tiser is to be used at the moment After vector augmentationthe fault detection problem is transformed into robust 119867
infin
filter problem design a robust fault detection filter for allallowable time-variant parameters external disturbance andpossible quantisations reduce the difference between fault
detection filterrsquos output and fault signal as small as possibleunder119867
infinframework The Lyapunov function is introduced
to ensure the asymptotically mean-square stability of faultdetection system and satisfies the corresponding119867
infindistur-
bance restraint conditions A sufficient condition is presentedto ensure the existence of robust fault detection filter inthe form of Linear Matrix Inequality (LMI) as well as thecalculation of fault detection filter parameters An examplelinearized from a vehicle system is used to show the efficiencyof the proposed method
Notations R119899 is the 119899-dimensional Euclidean space 1198972[0infin)
is the space of all square-summable vector functions over[0infin) 120575(119909 119910) is the Kronecker delta function for twointegers 119909 and 119910 if 119909 = 119910 120575(119909 119910) = 1 otherwise 120575(119909 119910) = 0Prob120591
119896= 119894means the probability of stochastic event 120591
119896= 119894
119868 stands for unit matrixE119909 is themathematical expectationof a stochastic variable 119909 |119909
119896| is the Euclidean norm vector of
119909119896 119909 = (suminfin
119896=0|119909119896|2
)12 is the 119897
2norm of 119909
119896 In symmetric
matrices ldquolowastrdquo represents a term that is induced by symmetryand diagsdot sdot sdot stands for a block-diagonal matrix
2 Problem Formulation
Consider a class of nonlinear NSs with ROQs
119909119896+1= 119860119896119909119896+ 119864119896119892 (119909119896) + 119861119896119908119896+119872119896119891119896
119910119896= 119862119909119896+ 119863119908
119896
(1)
where 119909119896isin R119899 is the system state vector 119908
119896isin 1198972[0infin) sub
R119899119908 is the external disturbance 119891119896isin R119899119891 is the fault signal
119860119896 119864119896 119861119896 and 119872
119896are time-variant matrices with suitable
dimensions and 119860119896= 119860 + Δ119860
119896 119864119896= 119864 + Δ119864
119896 119861119896= 119861 +
Δ119861119896 and119872
119896= 119872 + Δ119872
119896 where 119860 119864 119861 and119872 are known
constant matrices Δ119860119896 Δ119864119896 Δ119861119896 and Δ119872
119896are unknown
matrices with sector-bounded conditions given by
[Δ119860119896Δ119864119896Δ119861119896Δ119872119896] = 119873119865
119896[1198671119867211986731198674] (2)
where 119873 1198671 1198672 1198673 and 119867
4are known matrices 119865
119896
satisfies 119865119896119865119879
119896le 119868 119910
119896is the measurement signal under
ideal transmission conditions 119862 and 119863 are known constantmatrices119892(119909119896) is nonlinear vector function with sector restricted
conditions described as [31]
[119892 (119909119896) minus 1198791119909119896]119879
[119892 (119909119896) minus 1198792119909119896] le 0 forall119909
119896isin R119899119909 (3)
where1198791and1198792are known constantmatrices and119879 = 119879
1minus1198792
is symmetric and positive definite matrixConsidering the quantisation issue of NCs in this paper
we model the observing function as
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) 119892119894(119910119896) (4)
where 120591119896is a Bernoulli distributed stochastic variable with
a known distribution law Prob120591119896= 119894 = E120575(120591
119896 119894) = 119901
119894
sum119873
119894=1119901119894le 1 119910
119896is the measurement signal with quantisation
effect
Mathematical Problems in Engineering 3
For each 119894 (1 le 119894 le 119873) 119892119894(sdot) is a time-invariant
logarithmic quantiser which can be described by
119892119894(119910119896) = [119892
(1)
119894(119910(1)
119896) 119892(2)
119894(119910(2)
119896) sdot sdot sdot 119892
(119899119910)
119894(119910(119899119910)
119896)]
119879
(5)
where 119892(119895)119894(sdot) (1 le 119895 le 119899
119910) is a scalar quantisation function
with the quantisation levels described by
U(119895)
119894= plusmn119906
(119895)
119894119898| 119906(119895)
119894119898= (120588(119895)
119894)119898
119906(119895)
1198940 119898 = 0 plusmn1 plusmn2 cup 0
0 lt 120588(119895)
119894lt 1 119906
(119895)
1198940gt 0
(6)
Each of the quantisation levels corresponds to a seg-ment the data in one segment has the same quantisationoutput after being quantized The logarithmic quantisers aredescribed as
119892(119895)
119894(119910(119895)
119896) =
119906(119895)
119894119898
1
1 + 120575(119895)
119894
119906(119895)
119894119898lt119910(119895)
119896le
1
1 minus 120575(119895)
119894
119906(119895)
119894119898
0 119910(119895)
119896= 0
minus119892119895(minus119910(119895)
119896) 119910
(119895)
119896lt 0
(7)
where 120575(119895)119894
= (1 minus 120588(119895)
119894)(1 + 120588
(119895)
119894) Then we can obtain
120575(119895)
119894(119910(119895)
119896) = (1+Δ
(119895)
119894119896)119910(119895)
119896 where |Δ(119895)
119894119896| le 120575(119895)
119894 By definingΔ
119894119896=
diagΔ(1)119894119896 Δ
(119899119910)
119894119896 Δ119894= diag120575(1)
119894 120575
(119899119910)
119894 and 119865
119894119896=
Δ119894119896Δminus1
119894 we can obtain an unknown real-valued time-variant
matrix satisfying 119865119894119896119865119879
119894119896le 119868 Furthermore the measurements
with quantisation effects are given by
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119894119896Δ119894) 119910119896 (8)
Without loss of generality the subscript 119894 of 119865119894119896can be
ignored after vector augmentation and rewrite (8) as
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119910119896
=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119862119909119896+
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894)119863119908119896
(9)
Consider a full-order filter described as119909119896+1= 119866119909119896+ 119870119910119896
119903119896= 119871119909119896
(10)
where 119909119896is the state vector of the filter 119903
119896is output residual
119866 119870 and 119871 are the parameters to be determined
By defining 120578119896= [119909119879
119896119909119879
119896]119879 V119896= [119908119879
119896119908119879
119896]119879 and 119903
119896= 119903119896minus
119891119896 the augmented model is given by
120578119896+1= A119896120578119896+
119873
sum
119894=1
[120575 (120591119896 119894) minus 119901
119894]A0120578119896+E119896119892 (119885120578
119896) +B
119896V119896
119903119896= 119903119896minus 119891119896= C120578119896+DV119896
(11)where
A119896=[[
[
119860 0
119873
sum
119894=1
119901119894119870(119868 + 119865
119896Δ119894) 119862 119866
]]
]
+ [119873
0]119865119896[11986710] = A +N119865
119896H1
E119896= [119864
0] + [
119873
0]1198651198961198672= E +N119865
1198961198672
B119896=[[
[
119861 119872
119873
sum
119894=1
120575 (120591119896 119894) 119870 (119868 + 119865
119896Δ119894)119863 0
]]
]
+ [119873
0]119865119896[11986731198674] =B +N119865
119896H3
A0= [
0 0
119870 (119868 + 119865119896Δ119894) 119862 0
]
C = [0 119871] D = [0 minus119868] 119885 = [119868 0]
(12)
After the above augmentation the fault detection filterproblem can be formulated as design robust119867
infinfilter
The design of fault detection filter can be converted intodetermining a series of filter parameters119866119870 and 119871 tomakethe system satisfy two conditions
(1) under zero input condition system (11) is asymptoti-cally mean-square stable
(2) under zero initial condition for all possible nonzeroinputs V
119896 the output 119903
119896of (11) satisfies
E 1199032
lt 1205742
]2 (13)
where 120574 gt 0 is a prescribed119867infin
disturbance scalar
3 Main Results
31 Robust Fault Detection Filter Analysis
Theorem 1 For a given scalar 120574 gt 0 if there exist scalar 120575 andsymmetric and positive definite matrix 119875 = 119875119879 gt 0making thefollowing LMI holds
Φ =
[[[[
[
A119879119896119875A119896+
119873
sum
119894=1
119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1+C119879C A119879
119896119875E119896minus 120575T2
A119879119896119875B119896+C119879D
lowast E119879119896119875E119896minus 120575119868 E119879
119896119875B119896
lowast lowast B119879119896119875B119896minus 1205742
119868 +D119879D
]]]]
]
lt 0 (14)
4 Mathematical Problems in Engineering
whereT1= (TT1T2+ TT2T1)2 and T
2= minus(TT
1+ TT2)2 then
system (11) satisfies the aforementioned two conditions
Proof Consider the following Lyapunov function
119881119896= 120578119879
119896119875120578119896 (15)
ConsiderEsum119873119894=1[120575(120591119896 119894)minus119901
119894]2
= sum119873
119894=1119901119894(1minus119901119894) calculate
the difference of 119881119896under V
119896= 0 and its mathematical
expectation is
E Δ119881119896 = 120578119879
119896+1119875120578119896+1minus 120578119879
119896119875120578119896 (16)
where 120578119896+1= A119896120578119896+sum119873
119894=1119901119894(1minus119901119894)A0120578119896+E119896119892(119885120578119896) Formula
(3) can be rewritten as
[119909119896
119892(119909119896)]
119879
[T1
T2
T1198792119868]
119879
[119909119896
119892 (119909119896)] le 0 (17)
By defining 120585119896= [120578119879
119896119892119879
(119885120578119896)]119879 we can obtain
E Δ119881119896 le 120585119879
119896[Φ11Φ12
lowast Φ22
] 120585119896 (18)
where Φ11= A119879119896119875A119896+ sum119873
119894=1119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1
Φ12= A119879119896119875E119896minus 120575T2 and Φ
22= E119879119896119875E119896minus 120575119868
It can be inferred from LMI (14) that
[Φ11Φ12
lowast Φ22
] + [C119879C 0
0 0] lt 0 (19)
furthermore we can obtain EΔ119881119896 lt 0 From the above
proof it can be determined that the fault detection generalsystem (11) is asymptotically mean-square stable [32]
Let us consider the119867infin
disturbance restraint conditionsFor optional V
119896= 0 it can be inferred from (14) and (18) that
EΔ119881119896 + E119903119879
119896119903119896 minus 1205742EV119879119896V119896 = E120585119879
119896Φ120585119896 lt 0 where 120585
119896=
[120578119879
119896119892119879
(119885120578119896) V119879119896]119879 Then sum up the inequality from 0 toinfin
to obtaininfin
sum
119896=0
E 10038161003816100381610038161199031198961003816100381610038161003816
2
lt 1205742
infin
sum
119896=0
E 1003816100381610038161003816V1198961003816100381610038161003816
2
minus E 119881infin + E Δ119881
0 (20)
Since the Lyapunov function is positive-definite it is easy tosee that (13) holds under the initial conditionThis concludesthe proof
32 Robust Fault Detection Filter Design Before we discussthe robust fault detection filter design problem the followingtwo lemmas are useful in deriving the filter design results
Lemma 2 (see [33 34] (Schur complement)) Symmetricalmatrix
119878 = [1198781111987812
lowast 11987822
] (21)
is negative-definite when and only when
119887119900119905ℎ 11987811lt 0 119886119899119889 119878
22minus 119878119879
12119878minus1
1111987812lt 0 (22)
or
119887119900119905ℎ 11987822lt 0 119886119899119889 119878
11minus 119878119879
12119878minus1
2211987812lt 0 (23)
Lemma 3 (see [30]) If119872 = 119872119879119867 and 119864 are real matrices
and 119865 satisfies 119865119896119865119879
119896le 119868 then
119872+119867119865119864 + 119864119879
119865119879
119867119879
lt 0 (24)
holds when and only when there exists a constant 120576 gt 0making the following LMI holds
119872+1
120576119867119867119879
+ 120576119864119879
119864 lt 0 (25)
or
[
[
119872 119867 120576119864119879
lowast minus120576119868 0
lowast lowast minus120576119868
]
]
lt 0 (26)
Theorem4 For a given scalar 120574 gt 0 and the nonlinear discretetime-variant system characterized by (1)sim(3) there exists arobust fault detection filter in the form of (10) which makes thesystem asymptotically mean-square stable And the sufficientcondition satisfying the119867
infindisturbance restraint conditions is
that there exist real matrices 119877 = 119877119879 gt 0 119878 = 119878119879 gt 0 1198801 1198802
1198803and real scalar 120575 120576 gt 0 to make the following LMI holds
[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575T2
0 0 119860119879
119878 Ψ17
0 Ψ19119880119879
30 120576119867
119879
1
lowast Ψ22minus120575T2
0 0 119860119879
119878 Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119878 119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
119878 Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119878 119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (27)
Mathematical Problems in Engineering 5
where Ψ11= Ψ12= minus120575T
1minus 119878 Ψ
17= 119860119879
119877 + sum119873
119894=1119901119894(119868 +
119865119896Δ119894)119862119879
119880119879
2+ 119880119879
1 Ψ19= Ψ29= 120590(119868 + 119865
119896Δ119894)119862119879
119880119879
2 Ψ22=
minus120575T1minus119877Ψ
27= 119860119879
119877+sum119873
119894=1119901119894(119868+119865119896Δ119894)119862119879
119880119879
2Ψ47= 119861119879
119877+
sum119873
119894=1119901119894(119868+119865119896Δ119894)119863119879
119880119879
2 and 120590 = sum119873
119894=1radic119901119894(1 minus 119901
119894) If LMI (27)
holds then the robust fault detection filter parameters satisfyingthe above conditions can be calculated by
119866 = 119883minus1
121198801119878minus1
119884minus119879
12 119870 = 119883
minus1
121198802 119871 = 119880
3119878minus1
119884minus119879
12
(28)
and 11988312
and 11988412
here can be determined by 119868 minus 119877119878minus1 =11988312119884119879
12lt 0
Proof It can be inferred from Lemmas 2 and 3 that the LMI(14) equals
[[[[[[[[[[[[[[
[
minus119875 minus 120575T1minus120575T2
0 A119875 120590A1198790119875 C119879 0 120576H119879
1
lowast minus120575119868 0 E119875 0 0 0 120576119867119879
2
lowast lowast minus1205742
119868 B119875 0 D119879 0 120576H1198793
lowast lowast lowast minus119875 0 0 N 0
lowast lowast lowast lowast minus119875 0 0 0
lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]
]
lt 0 (29)
Define 119875 and 119875minus1 as the following blocks
119875 = [119877 119883
12
119883119879
1211988322
] 119875minus1
= [119878minus1
11988412
119884119879
1211988422
] (30)
where the dimension of a block is as the same asA
By defining
1198691= [119878minus1
119868
119884119879
120] 119869
2= [119868 119877
0 119883119879
12
] (31)
it can be see that 1198751198691= 1198692 11986911987911198751198691= 119869119879
11198692
By conducting contragradient transformation to thematrix of (29) with diag119869119879
1 119868 119868 119869
119879
1 119869119879
1 119868 119868 119868 we can obtain
[[[[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575119878minus119879T2
0 0 119878minus119879
119860119879
Ψ17
0 Ψ1911988412119871119879
0 120576119878minus119879
119867119879
1
lowast Ψ22
minus120575T2
0 0 119860119879
Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0 119878minus119879
119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (32)
where Ψ11= minus120575119878
minus119879T1119878minus1
minus 119878minus119879 Ψ12= minus120575119878
minus119879T1minus 119868
Ψ17= 119878minus119879
119860119879
119877 + 119878minus119879
sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119862119879
119870119879
119883119879
12 Ψ19=
120590119878minus119879
(119868 + 119865119896Δ119894)119862119879
119870119879
119883119879
12 Ψ22= minus120575T
1minus 119877 Ψ
27= 119860119879
119877 +
sum119873
119894=1119901119894(119868 +119865119896Δ119894)119862119879
119870119879
119883119879
12Ψ29= 120590(119868+119865
119896Δ119894)119862119879
119870119879
119883119879
12 and
Ψ47= 119861119879
119877 + sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119863119879
119870119879
119883119879
12
By introducing newmatrices1198801= 11988312119866119884119879
121198781198802=11988312119870
and 1198803= 119871119884119879
12119878 and conducting contragradient transforma-
tion to thematrix of (32) with diag119878 119868 119868 119868 119868 119878 119868 119878 119868 119868 119868 119868we can obtain LMI (27) It is to say that if the LMI (27) holdsthe fault detection augmentation system (11) is ensured to
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
2 Mathematical Problems in Engineering
years a generally adopted approach is transforming the resid-ual generation problem into optimization problem Design afault detection filter (residual generator) which makes thetransfer functionrsquos 119867
infinbound norm from disturbance to
residual as small as possible but the transfer functionrsquos 119867infin
bound norm from (or other bound norms) fault to residualas large as possible Some existing researches demonstratethat there exists a unified solving method among differentoptimal indices [18 19] After the residuals are generatedselect a suitable residual evaluation function and a thresholdvalue Residual evaluation functions and threshold values canbe different in form but usually we take residual evaluationfunction over than threshold value as the judgment of faultalarm Here is the description of fault detection problem thisis what we care about in this paper After fault detection isdone faults are located with suitable separation strategy levelof the faults is estimated by kinds of approaches superiorityof the diagnosis algorithm is evaluated This is the completefault diagnosis procedure
Most of the existing fault detection theories are set upbased on traditional control systems whose data transmis-sions are without delay or with ideal data dropout amongsystem components The appearance of NSs brings newchallenges to traditional fault detection theories and out ofthis the main problem is that the signals transmitted vianetwork became imperfect including delay data dropoutquantisation effect and other signal changes not expectedduring the design Compared to traditional control systemsthe difficulty of NSsrsquo fault detection mainly depends on theimperfection of observation signals By now most of theexisting research results of NSsrsquo diagnosis detection resultsfocus on dealingwith network induced delay and few of themfocus on data dropout multipacket transmission or signalquantisation [20ndash23] In [24ndash27] the authors studied systemrsquosstability and control problem with information coder In[28] Fu and Xie discussed the logarithmic quantiser anddesigned the corresponding state feedback controller Yang etal in [29] designed a networked filter based on logarithmicquantiser In [30] Li et al studied NSsrsquo fault diagnosisproblem whose residual errors have quantisation errors andthen designed fault diagnosis filters respectively based onstatic and dynamic coders But so far to the best of theauthorsrsquo knowledge there are very few related results in theliterature that focus on quantisation problems for NSs withrandomly occurring quantisations (ROQs) This inspires ourcurrent research
In this paper the robust fault detection and isolation(FDI) problem is investigated for a class of nonlinearNSswithROQs A time-variant system with unknown disturbancesand nonlinear terms is studiedThe parameters of the systemare time-variant and satisfy norm-bounded condition andall the quantisers are assumed to be in the logarithmic typewith different quantisation lawsOn the other hand Bernoullidistributed stochastic sequence is used to decide which quan-tiser is to be used at the moment After vector augmentationthe fault detection problem is transformed into robust 119867
infin
filter problem design a robust fault detection filter for allallowable time-variant parameters external disturbance andpossible quantisations reduce the difference between fault
detection filterrsquos output and fault signal as small as possibleunder119867
infinframework The Lyapunov function is introduced
to ensure the asymptotically mean-square stability of faultdetection system and satisfies the corresponding119867
infindistur-
bance restraint conditions A sufficient condition is presentedto ensure the existence of robust fault detection filter inthe form of Linear Matrix Inequality (LMI) as well as thecalculation of fault detection filter parameters An examplelinearized from a vehicle system is used to show the efficiencyof the proposed method
Notations R119899 is the 119899-dimensional Euclidean space 1198972[0infin)
is the space of all square-summable vector functions over[0infin) 120575(119909 119910) is the Kronecker delta function for twointegers 119909 and 119910 if 119909 = 119910 120575(119909 119910) = 1 otherwise 120575(119909 119910) = 0Prob120591
119896= 119894means the probability of stochastic event 120591
119896= 119894
119868 stands for unit matrixE119909 is themathematical expectationof a stochastic variable 119909 |119909
119896| is the Euclidean norm vector of
119909119896 119909 = (suminfin
119896=0|119909119896|2
)12 is the 119897
2norm of 119909
119896 In symmetric
matrices ldquolowastrdquo represents a term that is induced by symmetryand diagsdot sdot sdot stands for a block-diagonal matrix
2 Problem Formulation
Consider a class of nonlinear NSs with ROQs
119909119896+1= 119860119896119909119896+ 119864119896119892 (119909119896) + 119861119896119908119896+119872119896119891119896
119910119896= 119862119909119896+ 119863119908
119896
(1)
where 119909119896isin R119899 is the system state vector 119908
119896isin 1198972[0infin) sub
R119899119908 is the external disturbance 119891119896isin R119899119891 is the fault signal
119860119896 119864119896 119861119896 and 119872
119896are time-variant matrices with suitable
dimensions and 119860119896= 119860 + Δ119860
119896 119864119896= 119864 + Δ119864
119896 119861119896= 119861 +
Δ119861119896 and119872
119896= 119872 + Δ119872
119896 where 119860 119864 119861 and119872 are known
constant matrices Δ119860119896 Δ119864119896 Δ119861119896 and Δ119872
119896are unknown
matrices with sector-bounded conditions given by
[Δ119860119896Δ119864119896Δ119861119896Δ119872119896] = 119873119865
119896[1198671119867211986731198674] (2)
where 119873 1198671 1198672 1198673 and 119867
4are known matrices 119865
119896
satisfies 119865119896119865119879
119896le 119868 119910
119896is the measurement signal under
ideal transmission conditions 119862 and 119863 are known constantmatrices119892(119909119896) is nonlinear vector function with sector restricted
conditions described as [31]
[119892 (119909119896) minus 1198791119909119896]119879
[119892 (119909119896) minus 1198792119909119896] le 0 forall119909
119896isin R119899119909 (3)
where1198791and1198792are known constantmatrices and119879 = 119879
1minus1198792
is symmetric and positive definite matrixConsidering the quantisation issue of NCs in this paper
we model the observing function as
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) 119892119894(119910119896) (4)
where 120591119896is a Bernoulli distributed stochastic variable with
a known distribution law Prob120591119896= 119894 = E120575(120591
119896 119894) = 119901
119894
sum119873
119894=1119901119894le 1 119910
119896is the measurement signal with quantisation
effect
Mathematical Problems in Engineering 3
For each 119894 (1 le 119894 le 119873) 119892119894(sdot) is a time-invariant
logarithmic quantiser which can be described by
119892119894(119910119896) = [119892
(1)
119894(119910(1)
119896) 119892(2)
119894(119910(2)
119896) sdot sdot sdot 119892
(119899119910)
119894(119910(119899119910)
119896)]
119879
(5)
where 119892(119895)119894(sdot) (1 le 119895 le 119899
119910) is a scalar quantisation function
with the quantisation levels described by
U(119895)
119894= plusmn119906
(119895)
119894119898| 119906(119895)
119894119898= (120588(119895)
119894)119898
119906(119895)
1198940 119898 = 0 plusmn1 plusmn2 cup 0
0 lt 120588(119895)
119894lt 1 119906
(119895)
1198940gt 0
(6)
Each of the quantisation levels corresponds to a seg-ment the data in one segment has the same quantisationoutput after being quantized The logarithmic quantisers aredescribed as
119892(119895)
119894(119910(119895)
119896) =
119906(119895)
119894119898
1
1 + 120575(119895)
119894
119906(119895)
119894119898lt119910(119895)
119896le
1
1 minus 120575(119895)
119894
119906(119895)
119894119898
0 119910(119895)
119896= 0
minus119892119895(minus119910(119895)
119896) 119910
(119895)
119896lt 0
(7)
where 120575(119895)119894
= (1 minus 120588(119895)
119894)(1 + 120588
(119895)
119894) Then we can obtain
120575(119895)
119894(119910(119895)
119896) = (1+Δ
(119895)
119894119896)119910(119895)
119896 where |Δ(119895)
119894119896| le 120575(119895)
119894 By definingΔ
119894119896=
diagΔ(1)119894119896 Δ
(119899119910)
119894119896 Δ119894= diag120575(1)
119894 120575
(119899119910)
119894 and 119865
119894119896=
Δ119894119896Δminus1
119894 we can obtain an unknown real-valued time-variant
matrix satisfying 119865119894119896119865119879
119894119896le 119868 Furthermore the measurements
with quantisation effects are given by
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119894119896Δ119894) 119910119896 (8)
Without loss of generality the subscript 119894 of 119865119894119896can be
ignored after vector augmentation and rewrite (8) as
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119910119896
=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119862119909119896+
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894)119863119908119896
(9)
Consider a full-order filter described as119909119896+1= 119866119909119896+ 119870119910119896
119903119896= 119871119909119896
(10)
where 119909119896is the state vector of the filter 119903
119896is output residual
119866 119870 and 119871 are the parameters to be determined
By defining 120578119896= [119909119879
119896119909119879
119896]119879 V119896= [119908119879
119896119908119879
119896]119879 and 119903
119896= 119903119896minus
119891119896 the augmented model is given by
120578119896+1= A119896120578119896+
119873
sum
119894=1
[120575 (120591119896 119894) minus 119901
119894]A0120578119896+E119896119892 (119885120578
119896) +B
119896V119896
119903119896= 119903119896minus 119891119896= C120578119896+DV119896
(11)where
A119896=[[
[
119860 0
119873
sum
119894=1
119901119894119870(119868 + 119865
119896Δ119894) 119862 119866
]]
]
+ [119873
0]119865119896[11986710] = A +N119865
119896H1
E119896= [119864
0] + [
119873
0]1198651198961198672= E +N119865
1198961198672
B119896=[[
[
119861 119872
119873
sum
119894=1
120575 (120591119896 119894) 119870 (119868 + 119865
119896Δ119894)119863 0
]]
]
+ [119873
0]119865119896[11986731198674] =B +N119865
119896H3
A0= [
0 0
119870 (119868 + 119865119896Δ119894) 119862 0
]
C = [0 119871] D = [0 minus119868] 119885 = [119868 0]
(12)
After the above augmentation the fault detection filterproblem can be formulated as design robust119867
infinfilter
The design of fault detection filter can be converted intodetermining a series of filter parameters119866119870 and 119871 tomakethe system satisfy two conditions
(1) under zero input condition system (11) is asymptoti-cally mean-square stable
(2) under zero initial condition for all possible nonzeroinputs V
119896 the output 119903
119896of (11) satisfies
E 1199032
lt 1205742
]2 (13)
where 120574 gt 0 is a prescribed119867infin
disturbance scalar
3 Main Results
31 Robust Fault Detection Filter Analysis
Theorem 1 For a given scalar 120574 gt 0 if there exist scalar 120575 andsymmetric and positive definite matrix 119875 = 119875119879 gt 0making thefollowing LMI holds
Φ =
[[[[
[
A119879119896119875A119896+
119873
sum
119894=1
119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1+C119879C A119879
119896119875E119896minus 120575T2
A119879119896119875B119896+C119879D
lowast E119879119896119875E119896minus 120575119868 E119879
119896119875B119896
lowast lowast B119879119896119875B119896minus 1205742
119868 +D119879D
]]]]
]
lt 0 (14)
4 Mathematical Problems in Engineering
whereT1= (TT1T2+ TT2T1)2 and T
2= minus(TT
1+ TT2)2 then
system (11) satisfies the aforementioned two conditions
Proof Consider the following Lyapunov function
119881119896= 120578119879
119896119875120578119896 (15)
ConsiderEsum119873119894=1[120575(120591119896 119894)minus119901
119894]2
= sum119873
119894=1119901119894(1minus119901119894) calculate
the difference of 119881119896under V
119896= 0 and its mathematical
expectation is
E Δ119881119896 = 120578119879
119896+1119875120578119896+1minus 120578119879
119896119875120578119896 (16)
where 120578119896+1= A119896120578119896+sum119873
119894=1119901119894(1minus119901119894)A0120578119896+E119896119892(119885120578119896) Formula
(3) can be rewritten as
[119909119896
119892(119909119896)]
119879
[T1
T2
T1198792119868]
119879
[119909119896
119892 (119909119896)] le 0 (17)
By defining 120585119896= [120578119879
119896119892119879
(119885120578119896)]119879 we can obtain
E Δ119881119896 le 120585119879
119896[Φ11Φ12
lowast Φ22
] 120585119896 (18)
where Φ11= A119879119896119875A119896+ sum119873
119894=1119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1
Φ12= A119879119896119875E119896minus 120575T2 and Φ
22= E119879119896119875E119896minus 120575119868
It can be inferred from LMI (14) that
[Φ11Φ12
lowast Φ22
] + [C119879C 0
0 0] lt 0 (19)
furthermore we can obtain EΔ119881119896 lt 0 From the above
proof it can be determined that the fault detection generalsystem (11) is asymptotically mean-square stable [32]
Let us consider the119867infin
disturbance restraint conditionsFor optional V
119896= 0 it can be inferred from (14) and (18) that
EΔ119881119896 + E119903119879
119896119903119896 minus 1205742EV119879119896V119896 = E120585119879
119896Φ120585119896 lt 0 where 120585
119896=
[120578119879
119896119892119879
(119885120578119896) V119879119896]119879 Then sum up the inequality from 0 toinfin
to obtaininfin
sum
119896=0
E 10038161003816100381610038161199031198961003816100381610038161003816
2
lt 1205742
infin
sum
119896=0
E 1003816100381610038161003816V1198961003816100381610038161003816
2
minus E 119881infin + E Δ119881
0 (20)
Since the Lyapunov function is positive-definite it is easy tosee that (13) holds under the initial conditionThis concludesthe proof
32 Robust Fault Detection Filter Design Before we discussthe robust fault detection filter design problem the followingtwo lemmas are useful in deriving the filter design results
Lemma 2 (see [33 34] (Schur complement)) Symmetricalmatrix
119878 = [1198781111987812
lowast 11987822
] (21)
is negative-definite when and only when
119887119900119905ℎ 11987811lt 0 119886119899119889 119878
22minus 119878119879
12119878minus1
1111987812lt 0 (22)
or
119887119900119905ℎ 11987822lt 0 119886119899119889 119878
11minus 119878119879
12119878minus1
2211987812lt 0 (23)
Lemma 3 (see [30]) If119872 = 119872119879119867 and 119864 are real matrices
and 119865 satisfies 119865119896119865119879
119896le 119868 then
119872+119867119865119864 + 119864119879
119865119879
119867119879
lt 0 (24)
holds when and only when there exists a constant 120576 gt 0making the following LMI holds
119872+1
120576119867119867119879
+ 120576119864119879
119864 lt 0 (25)
or
[
[
119872 119867 120576119864119879
lowast minus120576119868 0
lowast lowast minus120576119868
]
]
lt 0 (26)
Theorem4 For a given scalar 120574 gt 0 and the nonlinear discretetime-variant system characterized by (1)sim(3) there exists arobust fault detection filter in the form of (10) which makes thesystem asymptotically mean-square stable And the sufficientcondition satisfying the119867
infindisturbance restraint conditions is
that there exist real matrices 119877 = 119877119879 gt 0 119878 = 119878119879 gt 0 1198801 1198802
1198803and real scalar 120575 120576 gt 0 to make the following LMI holds
[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575T2
0 0 119860119879
119878 Ψ17
0 Ψ19119880119879
30 120576119867
119879
1
lowast Ψ22minus120575T2
0 0 119860119879
119878 Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119878 119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
119878 Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119878 119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (27)
Mathematical Problems in Engineering 5
where Ψ11= Ψ12= minus120575T
1minus 119878 Ψ
17= 119860119879
119877 + sum119873
119894=1119901119894(119868 +
119865119896Δ119894)119862119879
119880119879
2+ 119880119879
1 Ψ19= Ψ29= 120590(119868 + 119865
119896Δ119894)119862119879
119880119879
2 Ψ22=
minus120575T1minus119877Ψ
27= 119860119879
119877+sum119873
119894=1119901119894(119868+119865119896Δ119894)119862119879
119880119879
2Ψ47= 119861119879
119877+
sum119873
119894=1119901119894(119868+119865119896Δ119894)119863119879
119880119879
2 and 120590 = sum119873
119894=1radic119901119894(1 minus 119901
119894) If LMI (27)
holds then the robust fault detection filter parameters satisfyingthe above conditions can be calculated by
119866 = 119883minus1
121198801119878minus1
119884minus119879
12 119870 = 119883
minus1
121198802 119871 = 119880
3119878minus1
119884minus119879
12
(28)
and 11988312
and 11988412
here can be determined by 119868 minus 119877119878minus1 =11988312119884119879
12lt 0
Proof It can be inferred from Lemmas 2 and 3 that the LMI(14) equals
[[[[[[[[[[[[[[
[
minus119875 minus 120575T1minus120575T2
0 A119875 120590A1198790119875 C119879 0 120576H119879
1
lowast minus120575119868 0 E119875 0 0 0 120576119867119879
2
lowast lowast minus1205742
119868 B119875 0 D119879 0 120576H1198793
lowast lowast lowast minus119875 0 0 N 0
lowast lowast lowast lowast minus119875 0 0 0
lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]
]
lt 0 (29)
Define 119875 and 119875minus1 as the following blocks
119875 = [119877 119883
12
119883119879
1211988322
] 119875minus1
= [119878minus1
11988412
119884119879
1211988422
] (30)
where the dimension of a block is as the same asA
By defining
1198691= [119878minus1
119868
119884119879
120] 119869
2= [119868 119877
0 119883119879
12
] (31)
it can be see that 1198751198691= 1198692 11986911987911198751198691= 119869119879
11198692
By conducting contragradient transformation to thematrix of (29) with diag119869119879
1 119868 119868 119869
119879
1 119869119879
1 119868 119868 119868 we can obtain
[[[[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575119878minus119879T2
0 0 119878minus119879
119860119879
Ψ17
0 Ψ1911988412119871119879
0 120576119878minus119879
119867119879
1
lowast Ψ22
minus120575T2
0 0 119860119879
Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0 119878minus119879
119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (32)
where Ψ11= minus120575119878
minus119879T1119878minus1
minus 119878minus119879 Ψ12= minus120575119878
minus119879T1minus 119868
Ψ17= 119878minus119879
119860119879
119877 + 119878minus119879
sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119862119879
119870119879
119883119879
12 Ψ19=
120590119878minus119879
(119868 + 119865119896Δ119894)119862119879
119870119879
119883119879
12 Ψ22= minus120575T
1minus 119877 Ψ
27= 119860119879
119877 +
sum119873
119894=1119901119894(119868 +119865119896Δ119894)119862119879
119870119879
119883119879
12Ψ29= 120590(119868+119865
119896Δ119894)119862119879
119870119879
119883119879
12 and
Ψ47= 119861119879
119877 + sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119863119879
119870119879
119883119879
12
By introducing newmatrices1198801= 11988312119866119884119879
121198781198802=11988312119870
and 1198803= 119871119884119879
12119878 and conducting contragradient transforma-
tion to thematrix of (32) with diag119878 119868 119868 119868 119868 119878 119868 119878 119868 119868 119868 119868we can obtain LMI (27) It is to say that if the LMI (27) holdsthe fault detection augmentation system (11) is ensured to
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 3
For each 119894 (1 le 119894 le 119873) 119892119894(sdot) is a time-invariant
logarithmic quantiser which can be described by
119892119894(119910119896) = [119892
(1)
119894(119910(1)
119896) 119892(2)
119894(119910(2)
119896) sdot sdot sdot 119892
(119899119910)
119894(119910(119899119910)
119896)]
119879
(5)
where 119892(119895)119894(sdot) (1 le 119895 le 119899
119910) is a scalar quantisation function
with the quantisation levels described by
U(119895)
119894= plusmn119906
(119895)
119894119898| 119906(119895)
119894119898= (120588(119895)
119894)119898
119906(119895)
1198940 119898 = 0 plusmn1 plusmn2 cup 0
0 lt 120588(119895)
119894lt 1 119906
(119895)
1198940gt 0
(6)
Each of the quantisation levels corresponds to a seg-ment the data in one segment has the same quantisationoutput after being quantized The logarithmic quantisers aredescribed as
119892(119895)
119894(119910(119895)
119896) =
119906(119895)
119894119898
1
1 + 120575(119895)
119894
119906(119895)
119894119898lt119910(119895)
119896le
1
1 minus 120575(119895)
119894
119906(119895)
119894119898
0 119910(119895)
119896= 0
minus119892119895(minus119910(119895)
119896) 119910
(119895)
119896lt 0
(7)
where 120575(119895)119894
= (1 minus 120588(119895)
119894)(1 + 120588
(119895)
119894) Then we can obtain
120575(119895)
119894(119910(119895)
119896) = (1+Δ
(119895)
119894119896)119910(119895)
119896 where |Δ(119895)
119894119896| le 120575(119895)
119894 By definingΔ
119894119896=
diagΔ(1)119894119896 Δ
(119899119910)
119894119896 Δ119894= diag120575(1)
119894 120575
(119899119910)
119894 and 119865
119894119896=
Δ119894119896Δminus1
119894 we can obtain an unknown real-valued time-variant
matrix satisfying 119865119894119896119865119879
119894119896le 119868 Furthermore the measurements
with quantisation effects are given by
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119894119896Δ119894) 119910119896 (8)
Without loss of generality the subscript 119894 of 119865119894119896can be
ignored after vector augmentation and rewrite (8) as
119910119896=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119910119896
=
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894) 119862119909119896+
119873
sum
119894=1
120575 (120591119896 119894) (119868 + 119865
119896Δ119894)119863119908119896
(9)
Consider a full-order filter described as119909119896+1= 119866119909119896+ 119870119910119896
119903119896= 119871119909119896
(10)
where 119909119896is the state vector of the filter 119903
119896is output residual
119866 119870 and 119871 are the parameters to be determined
By defining 120578119896= [119909119879
119896119909119879
119896]119879 V119896= [119908119879
119896119908119879
119896]119879 and 119903
119896= 119903119896minus
119891119896 the augmented model is given by
120578119896+1= A119896120578119896+
119873
sum
119894=1
[120575 (120591119896 119894) minus 119901
119894]A0120578119896+E119896119892 (119885120578
119896) +B
119896V119896
119903119896= 119903119896minus 119891119896= C120578119896+DV119896
(11)where
A119896=[[
[
119860 0
119873
sum
119894=1
119901119894119870(119868 + 119865
119896Δ119894) 119862 119866
]]
]
+ [119873
0]119865119896[11986710] = A +N119865
119896H1
E119896= [119864
0] + [
119873
0]1198651198961198672= E +N119865
1198961198672
B119896=[[
[
119861 119872
119873
sum
119894=1
120575 (120591119896 119894) 119870 (119868 + 119865
119896Δ119894)119863 0
]]
]
+ [119873
0]119865119896[11986731198674] =B +N119865
119896H3
A0= [
0 0
119870 (119868 + 119865119896Δ119894) 119862 0
]
C = [0 119871] D = [0 minus119868] 119885 = [119868 0]
(12)
After the above augmentation the fault detection filterproblem can be formulated as design robust119867
infinfilter
The design of fault detection filter can be converted intodetermining a series of filter parameters119866119870 and 119871 tomakethe system satisfy two conditions
(1) under zero input condition system (11) is asymptoti-cally mean-square stable
(2) under zero initial condition for all possible nonzeroinputs V
119896 the output 119903
119896of (11) satisfies
E 1199032
lt 1205742
]2 (13)
where 120574 gt 0 is a prescribed119867infin
disturbance scalar
3 Main Results
31 Robust Fault Detection Filter Analysis
Theorem 1 For a given scalar 120574 gt 0 if there exist scalar 120575 andsymmetric and positive definite matrix 119875 = 119875119879 gt 0making thefollowing LMI holds
Φ =
[[[[
[
A119879119896119875A119896+
119873
sum
119894=1
119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1+C119879C A119879
119896119875E119896minus 120575T2
A119879119896119875B119896+C119879D
lowast E119879119896119875E119896minus 120575119868 E119879
119896119875B119896
lowast lowast B119879119896119875B119896minus 1205742
119868 +D119879D
]]]]
]
lt 0 (14)
4 Mathematical Problems in Engineering
whereT1= (TT1T2+ TT2T1)2 and T
2= minus(TT
1+ TT2)2 then
system (11) satisfies the aforementioned two conditions
Proof Consider the following Lyapunov function
119881119896= 120578119879
119896119875120578119896 (15)
ConsiderEsum119873119894=1[120575(120591119896 119894)minus119901
119894]2
= sum119873
119894=1119901119894(1minus119901119894) calculate
the difference of 119881119896under V
119896= 0 and its mathematical
expectation is
E Δ119881119896 = 120578119879
119896+1119875120578119896+1minus 120578119879
119896119875120578119896 (16)
where 120578119896+1= A119896120578119896+sum119873
119894=1119901119894(1minus119901119894)A0120578119896+E119896119892(119885120578119896) Formula
(3) can be rewritten as
[119909119896
119892(119909119896)]
119879
[T1
T2
T1198792119868]
119879
[119909119896
119892 (119909119896)] le 0 (17)
By defining 120585119896= [120578119879
119896119892119879
(119885120578119896)]119879 we can obtain
E Δ119881119896 le 120585119879
119896[Φ11Φ12
lowast Φ22
] 120585119896 (18)
where Φ11= A119879119896119875A119896+ sum119873
119894=1119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1
Φ12= A119879119896119875E119896minus 120575T2 and Φ
22= E119879119896119875E119896minus 120575119868
It can be inferred from LMI (14) that
[Φ11Φ12
lowast Φ22
] + [C119879C 0
0 0] lt 0 (19)
furthermore we can obtain EΔ119881119896 lt 0 From the above
proof it can be determined that the fault detection generalsystem (11) is asymptotically mean-square stable [32]
Let us consider the119867infin
disturbance restraint conditionsFor optional V
119896= 0 it can be inferred from (14) and (18) that
EΔ119881119896 + E119903119879
119896119903119896 minus 1205742EV119879119896V119896 = E120585119879
119896Φ120585119896 lt 0 where 120585
119896=
[120578119879
119896119892119879
(119885120578119896) V119879119896]119879 Then sum up the inequality from 0 toinfin
to obtaininfin
sum
119896=0
E 10038161003816100381610038161199031198961003816100381610038161003816
2
lt 1205742
infin
sum
119896=0
E 1003816100381610038161003816V1198961003816100381610038161003816
2
minus E 119881infin + E Δ119881
0 (20)
Since the Lyapunov function is positive-definite it is easy tosee that (13) holds under the initial conditionThis concludesthe proof
32 Robust Fault Detection Filter Design Before we discussthe robust fault detection filter design problem the followingtwo lemmas are useful in deriving the filter design results
Lemma 2 (see [33 34] (Schur complement)) Symmetricalmatrix
119878 = [1198781111987812
lowast 11987822
] (21)
is negative-definite when and only when
119887119900119905ℎ 11987811lt 0 119886119899119889 119878
22minus 119878119879
12119878minus1
1111987812lt 0 (22)
or
119887119900119905ℎ 11987822lt 0 119886119899119889 119878
11minus 119878119879
12119878minus1
2211987812lt 0 (23)
Lemma 3 (see [30]) If119872 = 119872119879119867 and 119864 are real matrices
and 119865 satisfies 119865119896119865119879
119896le 119868 then
119872+119867119865119864 + 119864119879
119865119879
119867119879
lt 0 (24)
holds when and only when there exists a constant 120576 gt 0making the following LMI holds
119872+1
120576119867119867119879
+ 120576119864119879
119864 lt 0 (25)
or
[
[
119872 119867 120576119864119879
lowast minus120576119868 0
lowast lowast minus120576119868
]
]
lt 0 (26)
Theorem4 For a given scalar 120574 gt 0 and the nonlinear discretetime-variant system characterized by (1)sim(3) there exists arobust fault detection filter in the form of (10) which makes thesystem asymptotically mean-square stable And the sufficientcondition satisfying the119867
infindisturbance restraint conditions is
that there exist real matrices 119877 = 119877119879 gt 0 119878 = 119878119879 gt 0 1198801 1198802
1198803and real scalar 120575 120576 gt 0 to make the following LMI holds
[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575T2
0 0 119860119879
119878 Ψ17
0 Ψ19119880119879
30 120576119867
119879
1
lowast Ψ22minus120575T2
0 0 119860119879
119878 Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119878 119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
119878 Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119878 119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (27)
Mathematical Problems in Engineering 5
where Ψ11= Ψ12= minus120575T
1minus 119878 Ψ
17= 119860119879
119877 + sum119873
119894=1119901119894(119868 +
119865119896Δ119894)119862119879
119880119879
2+ 119880119879
1 Ψ19= Ψ29= 120590(119868 + 119865
119896Δ119894)119862119879
119880119879
2 Ψ22=
minus120575T1minus119877Ψ
27= 119860119879
119877+sum119873
119894=1119901119894(119868+119865119896Δ119894)119862119879
119880119879
2Ψ47= 119861119879
119877+
sum119873
119894=1119901119894(119868+119865119896Δ119894)119863119879
119880119879
2 and 120590 = sum119873
119894=1radic119901119894(1 minus 119901
119894) If LMI (27)
holds then the robust fault detection filter parameters satisfyingthe above conditions can be calculated by
119866 = 119883minus1
121198801119878minus1
119884minus119879
12 119870 = 119883
minus1
121198802 119871 = 119880
3119878minus1
119884minus119879
12
(28)
and 11988312
and 11988412
here can be determined by 119868 minus 119877119878minus1 =11988312119884119879
12lt 0
Proof It can be inferred from Lemmas 2 and 3 that the LMI(14) equals
[[[[[[[[[[[[[[
[
minus119875 minus 120575T1minus120575T2
0 A119875 120590A1198790119875 C119879 0 120576H119879
1
lowast minus120575119868 0 E119875 0 0 0 120576119867119879
2
lowast lowast minus1205742
119868 B119875 0 D119879 0 120576H1198793
lowast lowast lowast minus119875 0 0 N 0
lowast lowast lowast lowast minus119875 0 0 0
lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]
]
lt 0 (29)
Define 119875 and 119875minus1 as the following blocks
119875 = [119877 119883
12
119883119879
1211988322
] 119875minus1
= [119878minus1
11988412
119884119879
1211988422
] (30)
where the dimension of a block is as the same asA
By defining
1198691= [119878minus1
119868
119884119879
120] 119869
2= [119868 119877
0 119883119879
12
] (31)
it can be see that 1198751198691= 1198692 11986911987911198751198691= 119869119879
11198692
By conducting contragradient transformation to thematrix of (29) with diag119869119879
1 119868 119868 119869
119879
1 119869119879
1 119868 119868 119868 we can obtain
[[[[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575119878minus119879T2
0 0 119878minus119879
119860119879
Ψ17
0 Ψ1911988412119871119879
0 120576119878minus119879
119867119879
1
lowast Ψ22
minus120575T2
0 0 119860119879
Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0 119878minus119879
119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (32)
where Ψ11= minus120575119878
minus119879T1119878minus1
minus 119878minus119879 Ψ12= minus120575119878
minus119879T1minus 119868
Ψ17= 119878minus119879
119860119879
119877 + 119878minus119879
sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119862119879
119870119879
119883119879
12 Ψ19=
120590119878minus119879
(119868 + 119865119896Δ119894)119862119879
119870119879
119883119879
12 Ψ22= minus120575T
1minus 119877 Ψ
27= 119860119879
119877 +
sum119873
119894=1119901119894(119868 +119865119896Δ119894)119862119879
119870119879
119883119879
12Ψ29= 120590(119868+119865
119896Δ119894)119862119879
119870119879
119883119879
12 and
Ψ47= 119861119879
119877 + sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119863119879
119870119879
119883119879
12
By introducing newmatrices1198801= 11988312119866119884119879
121198781198802=11988312119870
and 1198803= 119871119884119879
12119878 and conducting contragradient transforma-
tion to thematrix of (32) with diag119878 119868 119868 119868 119868 119878 119868 119878 119868 119868 119868 119868we can obtain LMI (27) It is to say that if the LMI (27) holdsthe fault detection augmentation system (11) is ensured to
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
4 Mathematical Problems in Engineering
whereT1= (TT1T2+ TT2T1)2 and T
2= minus(TT
1+ TT2)2 then
system (11) satisfies the aforementioned two conditions
Proof Consider the following Lyapunov function
119881119896= 120578119879
119896119875120578119896 (15)
ConsiderEsum119873119894=1[120575(120591119896 119894)minus119901
119894]2
= sum119873
119894=1119901119894(1minus119901119894) calculate
the difference of 119881119896under V
119896= 0 and its mathematical
expectation is
E Δ119881119896 = 120578119879
119896+1119875120578119896+1minus 120578119879
119896119875120578119896 (16)
where 120578119896+1= A119896120578119896+sum119873
119894=1119901119894(1minus119901119894)A0120578119896+E119896119892(119885120578119896) Formula
(3) can be rewritten as
[119909119896
119892(119909119896)]
119879
[T1
T2
T1198792119868]
119879
[119909119896
119892 (119909119896)] le 0 (17)
By defining 120585119896= [120578119879
119896119892119879
(119885120578119896)]119879 we can obtain
E Δ119881119896 le 120585119879
119896[Φ11Φ12
lowast Φ22
] 120585119896 (18)
where Φ11= A119879119896119875A119896+ sum119873
119894=1119901119894(1 minus 119901
119894)A1198790119875A0minus 119875 minus 120575T
1
Φ12= A119879119896119875E119896minus 120575T2 and Φ
22= E119879119896119875E119896minus 120575119868
It can be inferred from LMI (14) that
[Φ11Φ12
lowast Φ22
] + [C119879C 0
0 0] lt 0 (19)
furthermore we can obtain EΔ119881119896 lt 0 From the above
proof it can be determined that the fault detection generalsystem (11) is asymptotically mean-square stable [32]
Let us consider the119867infin
disturbance restraint conditionsFor optional V
119896= 0 it can be inferred from (14) and (18) that
EΔ119881119896 + E119903119879
119896119903119896 minus 1205742EV119879119896V119896 = E120585119879
119896Φ120585119896 lt 0 where 120585
119896=
[120578119879
119896119892119879
(119885120578119896) V119879119896]119879 Then sum up the inequality from 0 toinfin
to obtaininfin
sum
119896=0
E 10038161003816100381610038161199031198961003816100381610038161003816
2
lt 1205742
infin
sum
119896=0
E 1003816100381610038161003816V1198961003816100381610038161003816
2
minus E 119881infin + E Δ119881
0 (20)
Since the Lyapunov function is positive-definite it is easy tosee that (13) holds under the initial conditionThis concludesthe proof
32 Robust Fault Detection Filter Design Before we discussthe robust fault detection filter design problem the followingtwo lemmas are useful in deriving the filter design results
Lemma 2 (see [33 34] (Schur complement)) Symmetricalmatrix
119878 = [1198781111987812
lowast 11987822
] (21)
is negative-definite when and only when
119887119900119905ℎ 11987811lt 0 119886119899119889 119878
22minus 119878119879
12119878minus1
1111987812lt 0 (22)
or
119887119900119905ℎ 11987822lt 0 119886119899119889 119878
11minus 119878119879
12119878minus1
2211987812lt 0 (23)
Lemma 3 (see [30]) If119872 = 119872119879119867 and 119864 are real matrices
and 119865 satisfies 119865119896119865119879
119896le 119868 then
119872+119867119865119864 + 119864119879
119865119879
119867119879
lt 0 (24)
holds when and only when there exists a constant 120576 gt 0making the following LMI holds
119872+1
120576119867119867119879
+ 120576119864119879
119864 lt 0 (25)
or
[
[
119872 119867 120576119864119879
lowast minus120576119868 0
lowast lowast minus120576119868
]
]
lt 0 (26)
Theorem4 For a given scalar 120574 gt 0 and the nonlinear discretetime-variant system characterized by (1)sim(3) there exists arobust fault detection filter in the form of (10) which makes thesystem asymptotically mean-square stable And the sufficientcondition satisfying the119867
infindisturbance restraint conditions is
that there exist real matrices 119877 = 119877119879 gt 0 119878 = 119878119879 gt 0 1198801 1198802
1198803and real scalar 120575 120576 gt 0 to make the following LMI holds
[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575T2
0 0 119860119879
119878 Ψ17
0 Ψ19119880119879
30 120576119867
119879
1
lowast Ψ22minus120575T2
0 0 119860119879
119878 Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119878 119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
119878 Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119878 119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878 minus119878 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (27)
Mathematical Problems in Engineering 5
where Ψ11= Ψ12= minus120575T
1minus 119878 Ψ
17= 119860119879
119877 + sum119873
119894=1119901119894(119868 +
119865119896Δ119894)119862119879
119880119879
2+ 119880119879
1 Ψ19= Ψ29= 120590(119868 + 119865
119896Δ119894)119862119879
119880119879
2 Ψ22=
minus120575T1minus119877Ψ
27= 119860119879
119877+sum119873
119894=1119901119894(119868+119865119896Δ119894)119862119879
119880119879
2Ψ47= 119861119879
119877+
sum119873
119894=1119901119894(119868+119865119896Δ119894)119863119879
119880119879
2 and 120590 = sum119873
119894=1radic119901119894(1 minus 119901
119894) If LMI (27)
holds then the robust fault detection filter parameters satisfyingthe above conditions can be calculated by
119866 = 119883minus1
121198801119878minus1
119884minus119879
12 119870 = 119883
minus1
121198802 119871 = 119880
3119878minus1
119884minus119879
12
(28)
and 11988312
and 11988412
here can be determined by 119868 minus 119877119878minus1 =11988312119884119879
12lt 0
Proof It can be inferred from Lemmas 2 and 3 that the LMI(14) equals
[[[[[[[[[[[[[[
[
minus119875 minus 120575T1minus120575T2
0 A119875 120590A1198790119875 C119879 0 120576H119879
1
lowast minus120575119868 0 E119875 0 0 0 120576119867119879
2
lowast lowast minus1205742
119868 B119875 0 D119879 0 120576H1198793
lowast lowast lowast minus119875 0 0 N 0
lowast lowast lowast lowast minus119875 0 0 0
lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]
]
lt 0 (29)
Define 119875 and 119875minus1 as the following blocks
119875 = [119877 119883
12
119883119879
1211988322
] 119875minus1
= [119878minus1
11988412
119884119879
1211988422
] (30)
where the dimension of a block is as the same asA
By defining
1198691= [119878minus1
119868
119884119879
120] 119869
2= [119868 119877
0 119883119879
12
] (31)
it can be see that 1198751198691= 1198692 11986911987911198751198691= 119869119879
11198692
By conducting contragradient transformation to thematrix of (29) with diag119869119879
1 119868 119868 119869
119879
1 119869119879
1 119868 119868 119868 we can obtain
[[[[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575119878minus119879T2
0 0 119878minus119879
119860119879
Ψ17
0 Ψ1911988412119871119879
0 120576119878minus119879
119867119879
1
lowast Ψ22
minus120575T2
0 0 119860119879
Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0 119878minus119879
119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (32)
where Ψ11= minus120575119878
minus119879T1119878minus1
minus 119878minus119879 Ψ12= minus120575119878
minus119879T1minus 119868
Ψ17= 119878minus119879
119860119879
119877 + 119878minus119879
sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119862119879
119870119879
119883119879
12 Ψ19=
120590119878minus119879
(119868 + 119865119896Δ119894)119862119879
119870119879
119883119879
12 Ψ22= minus120575T
1minus 119877 Ψ
27= 119860119879
119877 +
sum119873
119894=1119901119894(119868 +119865119896Δ119894)119862119879
119870119879
119883119879
12Ψ29= 120590(119868+119865
119896Δ119894)119862119879
119870119879
119883119879
12 and
Ψ47= 119861119879
119877 + sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119863119879
119870119879
119883119879
12
By introducing newmatrices1198801= 11988312119866119884119879
121198781198802=11988312119870
and 1198803= 119871119884119879
12119878 and conducting contragradient transforma-
tion to thematrix of (32) with diag119878 119868 119868 119868 119868 119878 119868 119878 119868 119868 119868 119868we can obtain LMI (27) It is to say that if the LMI (27) holdsthe fault detection augmentation system (11) is ensured to
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 5
where Ψ11= Ψ12= minus120575T
1minus 119878 Ψ
17= 119860119879
119877 + sum119873
119894=1119901119894(119868 +
119865119896Δ119894)119862119879
119880119879
2+ 119880119879
1 Ψ19= Ψ29= 120590(119868 + 119865
119896Δ119894)119862119879
119880119879
2 Ψ22=
minus120575T1minus119877Ψ
27= 119860119879
119877+sum119873
119894=1119901119894(119868+119865119896Δ119894)119862119879
119880119879
2Ψ47= 119861119879
119877+
sum119873
119894=1119901119894(119868+119865119896Δ119894)119863119879
119880119879
2 and 120590 = sum119873
119894=1radic119901119894(1 minus 119901
119894) If LMI (27)
holds then the robust fault detection filter parameters satisfyingthe above conditions can be calculated by
119866 = 119883minus1
121198801119878minus1
119884minus119879
12 119870 = 119883
minus1
121198802 119871 = 119880
3119878minus1
119884minus119879
12
(28)
and 11988312
and 11988412
here can be determined by 119868 minus 119877119878minus1 =11988312119884119879
12lt 0
Proof It can be inferred from Lemmas 2 and 3 that the LMI(14) equals
[[[[[[[[[[[[[[
[
minus119875 minus 120575T1minus120575T2
0 A119875 120590A1198790119875 C119879 0 120576H119879
1
lowast minus120575119868 0 E119875 0 0 0 120576119867119879
2
lowast lowast minus1205742
119868 B119875 0 D119879 0 120576H1198793
lowast lowast lowast minus119875 0 0 N 0
lowast lowast lowast lowast minus119875 0 0 0
lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]
]
lt 0 (29)
Define 119875 and 119875minus1 as the following blocks
119875 = [119877 119883
12
119883119879
1211988322
] 119875minus1
= [119878minus1
11988412
119884119879
1211988422
] (30)
where the dimension of a block is as the same asA
By defining
1198691= [119878minus1
119868
119884119879
120] 119869
2= [119868 119877
0 119883119879
12
] (31)
it can be see that 1198751198691= 1198692 11986911987911198751198691= 119869119879
11198692
By conducting contragradient transformation to thematrix of (29) with diag119869119879
1 119868 119868 119869
119879
1 119869119879
1 119868 119868 119868 we can obtain
[[[[[[[[[[[[[[[[[[[[[[[[[
[
Ψ11Ψ12minus120575119878minus119879T2
0 0 119878minus119879
119860119879
Ψ17
0 Ψ1911988412119871119879
0 120576119878minus119879
119867119879
1
lowast Ψ22
minus120575T2
0 0 119860119879
Ψ27
0 Ψ29
0 0 120576119867119879
1
lowast lowast minus120575119868 0 0 119864119879
119864119879
119877 0 0 0 0 120576119867119879
2
lowast lowast lowast minus1205742
119868 0 119861119879
Ψ47
0 0 0 0 120576119867119879
3
lowast lowast lowast lowast minus1205742
119868 119872119879
119872119879
119877 0 0 minus119868 0 120576119867119879
4
lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0 119878minus119879
119873 0
lowast lowast lowast lowast lowast lowast minus119877 0 0 0 119873 0
lowast lowast lowast lowast lowast lowast lowast minus119878minus119879
minus119868 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast minus119877 0 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast minus119868 0 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868 0
lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast lowast minus120576119868
]]]]]]]]]]]]]]]]]]]]]]]]]
]
lt 0 (32)
where Ψ11= minus120575119878
minus119879T1119878minus1
minus 119878minus119879 Ψ12= minus120575119878
minus119879T1minus 119868
Ψ17= 119878minus119879
119860119879
119877 + 119878minus119879
sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119862119879
119870119879
119883119879
12 Ψ19=
120590119878minus119879
(119868 + 119865119896Δ119894)119862119879
119870119879
119883119879
12 Ψ22= minus120575T
1minus 119877 Ψ
27= 119860119879
119877 +
sum119873
119894=1119901119894(119868 +119865119896Δ119894)119862119879
119870119879
119883119879
12Ψ29= 120590(119868+119865
119896Δ119894)119862119879
119870119879
119883119879
12 and
Ψ47= 119861119879
119877 + sum119873
119894=1119901119894(119868 + 119865
119896Δ119894)119863119879
119870119879
119883119879
12
By introducing newmatrices1198801= 11988312119866119884119879
121198781198802=11988312119870
and 1198803= 119871119884119879
12119878 and conducting contragradient transforma-
tion to thematrix of (32) with diag119878 119868 119868 119868 119868 119878 119868 119878 119868 119868 119868 119868we can obtain LMI (27) It is to say that if the LMI (27) holdsthe fault detection augmentation system (11) is ensured to
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
6 Mathematical Problems in Engineering
be asymptotically mean-square stable and satisfies the 119867infin
disturbance restraint conditions (13)On the other hand if LMI (27) holds then
[minus119878 minus119878
minus119878 minus119877] lt 0 (33)
and it can be inferred that 119878 minus 119877 lt 0 119878 gt 0 Since119875119875minus1
= 119868 then 119868 minus 119877119878minus1 = 11988312119884119879
12lt 0 By matrix singular
value decomposing two nonsingular square matrices 11988312
11988412
satisfying conditions can always be obtained thus thefilter parameters can be calculated by (28)This concludes theproof
33 Residual Evaluation Strategy After the fault detectionfilter is designed the residual evaluation function 119869
119896and
threshold value 119869th are introduced to evaluate NSs (whetherthere are faults in NSs)
119869119896=
119896
sum
119904=0
119903119879
119903119903119904
12
119869th = sup119908isin1198972119891=0
E 119869119896
(34)
The following inequalities are used to determine whetherthe fault occurs
119869119896gt 119869th 997904rArr Fault 997904rArr ALARM
119869119896le 119869th 997904rArr No Fault
(35)
4 Simulation Results
In this section in order to show the effectiveness of the pro-posed method a discrete-time linear system linearized froma vehicle system is employed with the following parameters
119860 = [
[
minus01 01 04
03 02 minus01
02 02 03
]
]
119864 = [
[
02 01 07
01 02 04
03 08 02
]
]
119861 = [
[
04
10825
40687
]
]
119872 = [
[
02
04
05
]
]
119873 = [
[
03
01
03
]
]
1198671= [02 01 05] 119867
2= [03 01 02]
1198673= [035] 119867
4= [02]
119862 = [02 01 05] 119863 = [03]
(36)
The nonlinear term 119892(119909119896) is defined as [119892
1(1199091119896) 0 119892
2(1199092119896)]119879
where 1198921(1199091119896) = 02119909
1119896times sin(119909
1119896) and 119892
2(1199092119896) = minus01119909
2119896times
sin(1199092119896) Then we can obtain that
1198791= [
[
025 0 0
0 0 0
0 0 0
]
]
1198792= [
[
0 0 0
0 0 0
0 0 015
]
]
(37)
The parameters of logarithmic quantisers are set as 119906(1)10=
119906(2)
10= 119906(3)
10= 119906(1)
20= 119906(2)
20= 119906(3)
20= 119906(1)
30= 119906(2)
30= 119906(3)
30= 1
0 5 10 15 20
0
5
10
15
20
25
Quantizer input
Qua
ntiz
er o
utpu
t
Quantizer 1Quantizer 2Quantizer 3
minus20 minus15 minus10 minus5minus25
minus20
minus15
minus10
minus5
Figure 1 Quantisation law of three quantisers
0 50 100 150 200 250 300 350 400 450 50005
1
15
2
25
3
35
Time (min)
Qua
ntiz
ater
sele
ctio
n sig
nal
Figure 2 Quantiser selection indicator
120588(1)
1= 120588(2)
1= 120588(3)
1= 06 120588(1)
2= 120588(2)
2= 120588(3)
2= 05 and 120588(1)
3=
120588(2)
3= 120588(3)
3= 04 as shown in Figure 1The distribution law of
the quantiser selection variable 120591119896is set as Prob120591
119896= 1 = 05
Prob120591119896= 2 = 035 and Prob120591
119896= 3 = 015 It is to say
that at a certain time the probability of quantiser 1 is 04the probability of quantiser 2 is 035 and the probability ofquantiser 3 is 025 which are illustrated in Figure 2
This paper aims to determine the parameters of the 119867infin
robust fault detection filter by using Theorem 4 With thehelp of mincx in Matlab [35] we can calculate the minimumvalue of 120574 as 120574lowast = radic1205742opt = 33321 where 120574
2
opt is the optimalsolution of the corresponding convex optimisation problems
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Mathematical Problems in Engineering 7
0 50 100 150 200 250 300 350 400 450 500
0
05
Time k
Time k0 50 100 150 200 250 300 350 400 450 500
005
1
minus05
wk
minus1
minus05
Fk
Figure 3 The time-domain simulations of 119908119896and 119865
119896
The parameters of the 119867infin
robust fault detection filter arecalculated as
119866 = [
[
minus01985 minus02718 minus0856
00005 03711 07378
minus00113 01834 06084
]
]
119870 = [
[
0044
minus01722
minus0031
]
]
119871 = [minus0029 minus00192 00945]
(38)
The unknown input 119908119896is defined as 05 times exp(minus11989610) times
(2119899119896minus 1) for 119896 = 0 1 500 where 119899
119896is uniformly
distributed in [minus1 1] The system parameter 119865119896is taken as
sin(1198962) as illustrated in Figure 3The fault signal 119891119896is taken
as 05 when 119896 isin [200 300] otherwise 119891119896= 0
As shown in Figure 4 (a) represents systemrsquos fault signal(b) represents the square of residuals 119903
119896 (c) represents the
output of residuals evaluation function 119869119896 By fixing the form
of 119908119896 we obtain 119869th = (1500)sum
500
119905=1119869500= 15997 times 10
minus4
after 500 times of simulation without fault As demonstratedin Figure 4(c) residuals evaluation function 119869
260= 15975 times
10minus4
lt 119869th lt 119869261 = 16462 times 10minus4 then it can be seen that the
fault is detected in the 61st step after it occurred
5 Conclusion
In this paper the analysis and design problem of an 119867infin
robust fault detection filter for a class of discrete-time NSswith ROQs have been investigated Because the bandwidthof the communication channel is limited signal must bequantised before it is transmitted via network ROQs areintroduced in the form of randomly occurring logarithmicquantisers which are transformed into randomly occurringuncertainties of system parameters By vector augmentingthe original fault detection problem is transformed intorobust119867
infinfilter problem Then a sufficient condition is pre-
sented to ensure the existence of robust fault detection filterwhich guarantees the asymptotically mean-square stability offault detection system and satisfies a certain119867
infindisturbance
0 50 100 150 200 250 300 350 400 450 500
0020406
minus02
fk
Time k
(a)
0 50 100 150 200 250 300 350 400 450 5000
05
1
r2 k
times10minus5
Time k
(b)
0 50 100 150 200 250 300 350 400 450 500024
J k
times10minus4
Time k
(c)
Figure 4 The time-domain simulations of 119891119896 1199032119896and 119869119896
restraint condition The simulation results demonstrate thatthis proposed approach is efficient for NSs with ROQs
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
References
[1] W Zhang M S Branicky and S M Phillips ldquoStability ofnetworked control systemsrdquo IEEE Control Systems Magazinevol 21 no 1 pp 84ndash99 2001
[2] M-Y Chow and Y Tipsuwan ldquoNetwork-based control systemsa tutorialrdquo in Proceedings of the 27th Annual Conference of theIEEE Industrial Electronics Society (IECON rsquo01) pp 1593ndash1602Denver Colo USA December 2001
[3] J P Hespanha P Naghshtabrizi and Y Xu ldquoA survey of recentresults in networked control systemsrdquo Proceedings of the IEEEvol 95 no 1 pp 138ndash172 2007
[4] L G Bushnell ldquoSpecial issue on networks and controlrdquo IEEEControl Systems Magazine vol 21 no 1 pp 132ndash143 2001
[5] P Antsaklis and J Baillieul ldquoSpecial issue on networked controlsystemsrdquo IEEE Transactions on Automatic Control vol 49 no4 pp 35ndash46 2004
[6] F-Y Wang D Liu S X Yang and L Li ldquoNetworking sensingand control for networked control systems architectures algo-rithms and applicationsrdquo IEEE Transactions on Systems Manand Cybernetics Part C Applications and Reviews vol 37 no 2pp 157ndash159 2007
[7] T J Koo and S Sastry ldquoSpecial issue on networked embeddedhybrid control systemsrdquo Asian Journal of Control vol 10 no 1pp 103ndash115 2008
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
8 Mathematical Problems in Engineering
[8] R M Murray K M Astrom S P Boyd R W Brockett andG Stein ldquoFuture directions in control in an information-richworldrdquo IEEE Control Systems vol 23 no 2 pp 20ndash33 2003
[9] D Liberzon ldquoOn stabilization of linear systems with limitedinformationrdquo IEEE Transactions on Automatic Control vol 48no 2 pp 304ndash307 2003
[10] S Wlmadssia K Saadaoui and M Benrejeb ldquoNew delay-dependent stability conditions for linear systems with delayrdquoSystem Science and Control Engineering vol 1 no 1 pp 2ndash112013
[11] L Zhang Y Shi T Chen and B Huang ldquoA new method forstabilization of networked control systemswith randomdelaysrdquoIEEE Transactions on Automatic Control vol 50 no 8 pp 1177ndash1181 2005
[12] D Yue Q-L Han and J Lam ldquoNetwork-based robust 119867infin
control of systems with uncertaintyrdquo Automatica vol 41 no 6pp 999ndash1007 2005
[13] F Yang Z Wang Y S Hung and M Gani ldquo119867infin
control fornetworked systems with random communication delaysrdquo IEEETransactions on Automatic Control vol 51 no 3 pp 511ndash5182006
[14] H Ahmada and T Namerikawa ldquoExtended Kalman filter-based mobile robot localization with intermittent measure-mentsrdquo System Science and Control Engineering vol 1 no 1 pp113ndash126 2013
[15] M Sahebsara T Chen and S L Shah ldquoOptimal1198672filtering in
networked control systems withmultiple packet dropoutrdquo IEEETransactions on Automatic Control vol 52 no 8 pp 1508ndash15132007
[16] H J Gao Y Zhao J Lam et al ldquo119867yen fuzzy filtering of nonlinearsystemswith intermittentmeasurementsrdquo IEEETransactions onSystems Fuzzy Systems vol 17 no 2 pp 291ndash300 2009
[17] J Chen and R J Patton Robust Model-Based Fault Diagnosis forDynamic Systems Kluwer Academic Boston Mass USA 1999
[18] X S Ding T Jeinsch M P Frank et al ldquoA unified approachto the optimization of fault detection systemsrdquo InternationalJournal of Adaptive Control and Signal Processing vol 14 no 7pp 725ndash745 2000
[19] L Qin X He and D H Zhou ldquoA survey of fault diagnosisfor swarm systemsrdquo Systems Science amp Control Engineering AnOpen Access Journal vol 2 no 1 pp 13ndash23 2014
[20] H Fang H Ye and M Zhong ldquoFault diagnosis of networkedcontrol systemsrdquo Annual Reviews in Control vol 31 no 1 pp55ndash68 2007
[21] Y-Q Wang H Ye and G-Z Wang ldquoRecent developmentof fault detection techniques for networked control systemsrdquoControl Theory and Applications vol 26 no 4 pp 400ndash4092009 (Chinese)
[22] X He Z Wang and D H Zhou ldquoRobust fault detectionfor networked systems with communication delay and datamissingrdquo Automatica vol 45 no 11 pp 2634ndash2639 2009
[23] W S Wong and R W Brockett ldquoSystems with finite commu-nication bandwidth constraints II Stabilization with limitedinformation feedbackrdquo IEEE Transactions on Automatic Con-trol vol 44 no 5 pp 1049ndash1053 1999
[24] X He Z Wang X Wang and D H Zhou ldquoNetworked strongtracking filteringwithmultiple packet dropouts algorithms andapplicationsrdquo IEEE Transactions on Industrial Electronics vol61 no 3 pp 1454ndash1463 2014
[25] R W Brockett and D Liberzon ldquoQuantized feedback stabiliza-tion of linear systemsrdquo IEEE Transactions on Automatic Controlvol 45 no 7 pp 1279ndash1289 2000
[26] S Tatikonda and S Mitter ldquoControl under communicationconstraintsrdquo IEEE Transactions on Automatic Control vol 49no 7 pp 1056ndash1068 2004
[27] X He Z Wang Y Liu and D H Zhou ldquoLeast-squaresfault detection and diagnosis for networked sensing systemsusing a direct state estimation approachrdquo IEEE Transactions onIndustrial Informatics vol 9 no 3 pp 1670ndash1679 2013
[28] M Fu and L Xie ldquoThe sector bound approach to quantizedfeedback controlrdquo IEEE Transactions on Automatic Control vol50 no 11 pp 1698ndash1711 2005
[29] Y H Yang W Wang and F W Yang ldquoQuantified Hinfinfilter for
discrete-timeMIMO systemsrdquoControl andDecision vol 24 no6 pp 932ndash936 2009 (Chinese)
[30] W Li P Zhang S X Ding and O Bredtmann ldquoFault detectionover noisy wireless channelsrdquo in Proceedings of the 46th IEEEConference on Decision and Control (CDC rsquo07) pp 5050ndash5055New Orleans La USA December 2007
[31] Q-L Han ldquoAbsolute stability of time-delay systemswith sector-bounded nonlinearityrdquo Automatica vol 41 no 12 pp 2171ndash2176 2005
[32] X Mao Differential Equations and Applications HorwoodPublishing Chichester UK 1997
[33] S Boyd L El Ghaoui E Feron and V Balakrishnan Lin-ear Matrix Inequalities in System and Control Theory SIAMPhiladelphia Pa USA 1994
[34] K M Grigoriadis and J T Watson Jr ldquoReduced-order119867infinand
1198712-119871infin
filtering via linear matrix inequalitiesrdquo IEEE Transac-tions on Aerospace and Electronic Systems vol 33 no 4 pp1326ndash1338 1997
[35] P Gahinet A Nemirovski J A Laub et al LMI ControlToolbox-for Use with Matlab The Math Works Natick MassUSA 1995
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical Problems in Engineering
Hindawi Publishing Corporationhttpwwwhindawicom
Differential EquationsInternational Journal of
Volume 2014
Applied MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Probability and StatisticsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Mathematical PhysicsAdvances in
Complex AnalysisJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OptimizationJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
CombinatoricsHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Operations ResearchAdvances in
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Function Spaces
Abstract and Applied AnalysisHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of Mathematics and Mathematical Sciences
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Algebra
Discrete Dynamics in Nature and Society
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Decision SciencesAdvances in
Discrete MathematicsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom
Volume 2014 Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Stochastic AnalysisInternational Journal of