STARTING CURRENT REGULATION OF INDUCTION MOTOR … · Comparatively star/delta starting is the most preferable ... motor smooth starting is a dynamic process and we ... the rst place

STARTING CURRENT REGULATIONOF INDUCTION MOTOR THROUGH

(NPSO)

A.Chandra Sekhar1, Dr.N. Bhoopal2

Kaviti Dilip Kumar3 ,Madhavi4

Asst.,Prof1,Prof.Head of the department 2

B V Raju Institute Of TechnologyNarsapur ,Medak(Dt)

[email protected]@[email protected]

July 20, 2018

Abstract

The proposal in this research paper is starting currentregulation for 3-φ induction motor through feedbackcontroller. The challenging task in induction motor isstarting current regulation. To ensure smooth starting ofinduction motor we need a device that can control thecurrent inrush. The innovation in this paper isprobabilistic based Particle Swarm Optimization for the PIcontroller. A standard PSO is modified by appropriateunification of probabilistic method, Elitism, andpopulation adjustment taken from existing biologicalenhancement approaches like Genetic Algorithm (GA),Bees GA, Artificial immune system (AIS) etc. In theupdated version of PSO, the randomly generation ofparticles will be based on the probability, so that with lessiteration we can achieve the optimized values of PIcontroller. Some standard test functions are used to test

1

International Journal of Pure and Applied MathematicsVolume 120 No. 6 2018, 8187-8207ISSN: 1314-3395 (on-line version)url: http://www.acadpubl.eu/hub/Special Issue http://www.acadpubl.eu/hub/

8187

the updated PSO. Once the algorithm is successful in thetest cases, then it is used for the soft starting of inductionmotor. Another crux of this paper is well tunedProportional integral (PI) controller for closed loopoperation. These testing results show cased at the time ofsmooth starting. The experimental results also depictedalong with simulation.

Keywords:Induction motor, ,updated Particle swarmoptimization technique (MPSO), Probabilistic technique,Elitism technique, population technique, GeneticAlgorithm, Bees (GA), Artificial Immune System (AIS).

NOMENCLATURE:

Vds = stator voltage along direct axis.Vqs = stator voltage along quadrature axis.ids = stator current along direct axis.iqs = stator current along quadrature.idr = rotor current along quadrature axis.iqr = Quadrature axis vector component of rotor current.ir,iy,ib = Per phase current of induction motor.rs = stator resistancerr = rotor resistanceLm = Magnetizing inductanceLs = stator inductanceLr = rotor inductanceωr = Angular speedω = supply voltage angular frequencyTe = Electromagnetic torqueTL = Load torqueP = Number of polesJ = motor moment of inertiaα = firing angle representationI = hot line currentI∗ = reference rated currente = Feedback error from motorαo = Initial triggering angleKp = Induction motor constant proportional gainKI = Induction motor constant integral gainC1, C2 = Updated PSO positive constants

2

International Journal of Pure and Applied Mathematics Special Issue

8188

W = Updated PSO inertial weightPrg = random generation of particle by probability.

1 INTRODUCTION

The basic methods for starting an induction motor is DOLstarting, Electromechanical reduced voltage starting, solid statereduced voltage starting and variable frequency drive (VFD)starting. Direct online starting is the most economical way, butthe poking for technology is drawing of high starting current.Another way of starting of an induction motor iselectromechanical reduced voltage, which is being skeleton intoauto transformer starting, star/delta starting and resistor/reactorstarting. Comparatively star/delta starting is the most preferableone as it is economical, less space consuming and having lowpower losses. But this method can be used only for normallydelta- connected star provided with six leads.

Generally starting methods need a mechanical switch or contactand have several disadvantages like, need a time to timeexamination and maintenance, non-synchronous switching ofmotor phases(RYB) to the mains. Large amount of switchingfurther leads to breakdown of moving parts, etc.

Nowadays sophisticated electrical technology has been evolved.The availability of these high-power devices demands sophistoperation of drives. There by the innovation heading towardsbio-inspired algorithms like Particle swarm optimization (PSO).The testing results ensure PSO is the most useful techniques forsolving the global developmental problems. This method wasprimarily designed by J. Kennedy as an optimization method in1995 and later on this technique has been proved to be one of thevaluable techniques for solving the global optimization problems.

With SCR based voltage regulator operating in closed loopmechanism, makes the smooth starting of induction motor byvarying the firing angle. the major essence in the ac voltageregulator with induction motor in closed loop is that we can vary

3


8189

the voltage across the thyristor. Whereas at past works weregrounded to current limit starting, the starting torque profile ofinduction motor for the better improvement was addressed later.By taking the voltage and power factor as feedback for a softstarter there is a development in the starting torque of themachine and artificial neural network (ANN) is also used for thegeneration of thyristor firing angles for smooth starting of themachine. Soft starters are also useful for saving of power.

The mitigation of starting current spikes resulted in extensiveusage of soft starter even in large industries. But thedifferentiation proposed in this paper is soft starter works inclosed loop mechanism. Closed loop control methods are normallyobtained by iteratively simulating the induction motor model.

Innovation in this paper is induction motor smooth starting isemployed by applying small signal model. Basically inductionmotor smooth starting is a dynamic process and we cant beexplained that with steady state equivalent circuit. New techniqueis proposed updated PSO technique in which it searches the finestvalues of PI controller for induction motor smooth starting.

The main function of updated particle swarm optimization iswhen the set of particles are taken in order to depict the tunedparticle. When the number of particles iterates for certain numberof times then after some iterations the best particle would occupythe first place and remaining particles occupies with theirrespective position based upon their functionality behavior. Againthose particles which have come from the first operation are againgoes to operation and in that finds out the best particle and thoseparticles are applied to the system for better performance. Whenthose particles (poles and zeros) are plotted on the systemstability graph, the main concentration is on the particles whichare nearer to the origin because those particles only effects thesystem performance compared to the particles far away from theorigin.

Due to some of the limitations PSO was having some drawbacksand they can be overcome by Time varying acceleration

4


8190

coefficients (TVAC), particle swarm optimizer with mutation andTime varying acceleration coefficients (MPSO-TVAC) and self-organizing hierarchal particle swarm optimization with timevarying acceleration coefficients (HPSO- TVAC). A collegialapproach to standard PSO can be achieved by the improvedparticle swarm optimization methods. The updated version of psosignifies the smooth performance which is depicted in simulationand hardware results. This paper mainly represents updatedstructure of PSO, whereas the concepts of probability, Elitism andpopulation are taken from some of the biologically inspiredalgorithms like genetic algorithm (GA) Ant colony optimization(ACO), Artificial immune system (AIS) etc. whereas ACO dealswith probabilistic approaches. Genetic algorithm uses one or twobest chromosomes in elitism and they will retain withoutundergoing to some selection, crossover and mutation.

Due to this advanced techniques availability the new PSO couldproduce its result in less number of iterations. On this process thenew PSO is tested in different test functions and find out the bestone and verified it. Due to this characteristic functions the novelalgorithm is used for smooth starting of induction motor.

2 DEVELOPMENT OF MODIFIED

PARTICLE SWARM

OPTIMIZATION

The primitive equations which control the movement of theparticle in standard in particle swarm optimization are givenbelow: the predicted position of ith particle of the swarm andspeed of the particle at (k+1)th iteration are defined as:

V k+1ij = w.V k

ij + C1rand() × [pbestkij −Xkij] + C2rand() × [gbestkj −

Xkij].....(1)

Xk+1ij = Xk

ij + V k+1ij .......(2)

Where i =1, 2, 3., p is the particle index, w is the inertial weight,and c1 and c2 are positive constant, called the acceleration

5


8191

constants r () and is a random number, systematically allocatedwithin the interval [0, 1] j is the size (dimension) of the problem.

In PSO, after certain iterative operations the newly producedparticle is within the direction of vector sum of the position of theparticle which best fits within that position and its previous bestposition of the particle.

In PSO, the importance of random function is it gives a result ofprobabilistic particle movement. But, the quantum ofprobabilistic approach is limited because the speed of the particleis limited which is defined on equation (1) is a resultant of particlebest and global best. Due to these limitations conventionalapproach of PSO is not suitable for finding the solution spacequickly and efficiently. These can also results in reduced rate ofconvergence and performance of PSO.

In the modified method, the particles of population size, p isselected in a random manner and in those pbest and gbest areidentified easily. In conventional approach the velocity andposition of an each particle is updated for every iteration in PSO.But, in modified version of PSO a random number, pi , 0 ¡ pi ¡1 isgenerated each time and compared with the randomly generatedparticle prg . For the condition pi ¿=prg , in this the best particleis retained to first place and remaining (p-1) particles are replacedwith randomly produced particles. For another condition, this iswhen pi¡prg , for this the originally generated population isretained and the loop will continuously iterate.

The pseudo code for updated PSO is given below:

Firstly the three parameters, like Max iteration, population size Pand probability of random generation Prg would be initialized.

P particles which are taken randomly are allowed into the solutionspace.

In those randomly generated particles, figure out the objectivefunction value, for each particle at the end of the problem.

6


8192

While after randomly generating the particles. For each particleinitialize the present location as Pbest (particle best).

Find out the gbest.

Upgrade the current position and speed of each particle by usingequations (1) and (2).

For (i¡Max iteration) Generate a random number,

The particle which is retained that particle having the gbest andremaining particles all are deleted; generate (P-1) particlesrandomly.ElseGo to step 8

Upgrade (Renovate) Pbest and gbest of each particle.

i = i+1End

7


8193

In order to calculate the capability of the suggested algorithm,considerable simulations were carried out on a number of testfunctions. By applying these test functions the performance alsoincluded in the figure. The curves which we got are taken bydoing, after averaging of 20 trail ru.

For each train run what we had used for initial population is usedfor each technique for easiness of comparison. Whatever theresults that had been obtained after trailing each test function islisted in the table. From the above listed table and from theperformance graph we can get in to one conclusion that, theproposed algorithm is the best standard PSO compared toprevious methods. By using this technique we get quality output

8


8194

results and we get output in less number of iterations. Thisclearly determines the excellence of standard PSO.

The parameters what we had obtained from standard PSO istabulated in table 2.

The results obtained by using standard PSO are shown in thefigurative graph. In the improved version of PSO it is easy tocalculate the test functions and performance graphs of thesemodified techniques.

3 APPLICATION OF NEW PSO

FOR FEEDBACK CONTROLLER

DESIGN

Designing of induction soft starting

The representational diagram of induction motor soft starting isshown in the figure. Here motor design is explained by the fifthorder differential equation and it’s written as:

Where p=d/dt

The electromagnetic torque is given by:

Te=32∗ P

2Lm(idriqs − iqrids)......(4)

The speed of the rotor is given by:

9


8195

Pωr = Te−TL

J........(5

The voltages at stator are Vqs and Vds in equation(3) are given by

[VqsVds

]=

[1 0 00 − 1√

31√3

]VrsVbsVys

Where Vrs, Vbs and Vys are motor terminal voltages and arederived below:

We can vary the speed of the induction motor by varying thestator side voltage. It can be done by using an AC voltageregulator, by changing the parameter α called as firing angle,which is connected in between the induction motor and supply.

Voltage components of stator side and with its instantaneousphase voltages can be written as:

Vr = Vm sin(ωt)......(7)Vy = Vm sin(ωt− 2π/3).......(8)Vb = Vm sin(ωt+ 2π/3)......(9)

Here ω is the angular frequency in rad/s of sinusoidal supplyvoltage.

The motor phase voltage between the SCR voltage controller andsupply phase voltage is given by:

Vr = Vm sin(ωt), when ir 6= zero and ωt ¡ α= er when ir 6= zero and ωt < α.......(10)=Vm(sinωt), whenir 6= zeroandωt ≥ α

Similarly:

Vy = Vm(sinωt− 2π/3), wheniy 6= zeroandωt < α= eywheniy = zeroandωt < α........(11)= Vm(sinωt− 2π/3), wheniy 6= zeroandωt ≥ α

Vb = Vm(sinωt+ 2π3), whenib 6= zeroandω < α

10


8196

= ebwhenib = zeroandωt < α..........(12)=(sinωt+ 2π/3), whenib 6= zeroandωt ≥ α

While controlling the SCR voltage controller with three phases R,Y, B

There is another parameter called neutral shit voltage and is givenby:

Vsn = 13(Vr + Vy + Vb)..........(13)

Thus the three phase forcing voltages at the induction motorterminal are given by:

Vrs = (Vr − Vsn)Vys = (Vy − Vsn).......(14)Vbs = (Vb − Vsn)

The emf induced in each phase is written as:iriyib

=

√23

1 0

−12

√32

−12−√32

[idsiqs

].......(15)

At the d-q axis the back emf induced in the motor terminals are:

[edreqr

]= ωr

[lrliqrlrlidr

] [ψq

ψd

]......(16)

Where;

φlower ≤ φ ≤ φupperand0 ≤∝≤ πinwhich[ψd

ψq

]=

[ψdr −lrliqrψqr −lrlidr

]......(17)

The rotor flux linkages in the direct and quadrant axis ψdrψqrare:

11


8197

[ψdr

ψqr

]=

[Lr Lm 0 00 0 Lr Lm

]

idsiqsidr

iqr

......(18)

The controlling of induction motor design is by using SCR firingangle α and load torque TL . At the initial stage motor is startedat no-load i.e.

TL= 0 and motor current I is calculated through the model andthat current is compared with the motor rotor current I∗.

The main function here is to we can vary the motor current up toits rated value during the initial time.For this controlling action normally proportional integral (PI)controller is suggested. Thus we can see the controlling action bythe variation of parameter ∆α .α0 The value of Tells us thestarting SCR firing angle and is acceptably chosen.

4 PROBLEM DESIGNING

The input given to the motor smooth starting is Kp and Kt value.The motor current is made constant throughout the starting. Theerror ie= I-I∗ is minimized by optimization, this is started asoptimized problem and is given as minimized error

F (φ) =∑ts

0 (e(t)2)

This is subjectedTs is starting internalφ = {Kp, Kt} Is the controller structure

The performance graph:

A Specific program in mat lab is designed for controlleridentification for newer PSO. The inputs to the proposed PSO areC1, C2, inertia factor, and probability of movement, populationsize and no of iterations. These all inputs are normally

12


8198

experimented and experienced to obtained the best solution set,the parameter between for the new PSO are listed in the table

The convergence graph for the conventional PSO and newer PSOare obtained. The graph can be plotted after giving 20 trail runs.The each trail is perfectly scrutinized for the better output.

The convergence graph clearly shows the comparison between theConventional and newer PSO. The better result obtained can eeasily identified. The result obtained after the 30th iteration withthe control constants are:

Kp =7.27 and Kt =15.8851

The controller constants obtained from standard PSO are

Kp =9.92 and Kt=6.68

Simulated and measured results:

This section will clearly describe about the experimental resultcarried out on a laboratory motor with advanced PSO withconventional and new PSO. The soft starting of motor is firstsimulated with help of motor dynamic equation and thus, verifiedwith proto type motor which is available in laboratory. The softstarting dynamic model of motor is shown in figure 3 and theexperimental setup shown in figure 5.

A 3-phase AC voltage regulator is employed to a power converter.In order to adjust the stator voltage more the firing angle isvaried. The AC voltage controller consists of six SCR’s labeledasT1, T1’, T2, T2’, T3 and T3’. At a delay angle of α and π + αSCRs T1 and T1’ are triggered respectively with reference to thezero crossing of R-phase voltage. At delay angles of (2π/3)+α and(2π/3)+(π+α) the SCRs T2 and T2’ are triggered respectively. Inthe same way at the delay angles of (4π/3) + α and (4π/3) +(π+α) the SCRs T3 and T3’ are triggered respectively.

Microcontroller is the heart of any closed loop system and

13


8199

microcontroller works as PI controller and also as a firing pulsegenerator. The reference voltage of R- phase VRN is steppeddown and by using zero crossing detection (ZCD) it is convertedinto digital pulse and then this pulse is connected to the pin RC0of the microcontroller. The status of this pin is sensed by themicrocontroller at every second, it starts producing six firingpulses through RB0 TO RB5 with a delay angle, which iscomputed by PI controller which is realized by using themicrocontroller once the zero crossing signal will get appeared.

Extensive simulations and experimentations are being carried outin order to validate soft starting scheme. The motor currenttransient during starting is shown in figure 6(a). The soft startingwith the new PSO based PI current regulator is excellent withzero overshoot and least ripple is suggested by the nature ofvariation of motor current. The soft starting is also performedusing PI controller tuned through standard PSO for comparison.Figure 6(b) shows the current transient with this scheme.

The motor current will have least ripple but it will have largeovershoot. The figure 4 shows the convergence graph whichdescribes difference in transient nature of motor current can begauged.

The objective function value is 48.36 at the end of 30 iterationswith a new PSO. The value is 98.30 with conventional PSO. Thedifference in the quality gives raise to different transient motorcurrent variation with the new PSO and conventional PSO.

Circuit of Closed loop control of induction motor

14


8200

Schematic diagram of the microcontroller based soft starter forinduction motor

15


8201

TABLE 1STANDARD TEST FUNCTION RESULTS

TABEL IIPARAMETERS OF ALGORITHMS

16


8202

TABLE IIIUPDATED PSO PARAMETERS

TABEL IVMEASURED VALUE OF KEY PARAMETERS

5 SIMULATED AND HARDWARE

RESULTS

Fig:1(speed RMS current)

17


8203

Fig:2(instantaneous voltage current)

Fig:3(instantaneous voltage current after soft startingcompletion)

18


8204

Fig:4(instantaneous voltage current in the entire process of softstarting.)

6 CONCLUSION

Closed loop mechanism for a induction motor drive namedsmooth starter is the focused area.

The soft starter fed 3-φ induction motor drive with closed loopoperation is first designed in MATLAB software .

Using updated version PSO results of the feedback controllerconstants identification for enhanced starting dynamics. Theconvergence values of the new PSO are tested on standardfunctions and then used for estimating optimal structure of motorcurrent regulator. The complete equipment is first of all tested inthe MATLAB and then finally implemented in the laboratory.

The main essence of this paper is tuned PI controller constantsobtained through Probalistic PSO and hard ware implementationthrough low cost pic microcontroller(16f876) feedback controllervalidation through hardware.

19


8205

References

[1] 1. W.S. Wood, F. Flynn, A. Shanmughasundaram. Transienttorques in induction motors due to switching of the supply.Proc. IEE, vol. 112, no. 7, pp. 1348-1354, 1965.Show ContextView Article Full Text: PDF (840KB)

[2] W.J. Lukitsch.Soft start vs AC drives Understand thedifferences. Conference Proceedings of IEEE Textile Fiber andFilm Industry. pp. 1-5, 1999.Show Context View Article FullText: PDF (347KB)

[3] H. Xiaohui, S. Yuhui, R. Eberhart.Recent advances in particleswarm. Proc. Congr. Evolut. Comput., vol. 1, pp. 90-97, 2004.Show Context View Article Full Text: PDF (498KB)

[4] J. Kennedy, R. Eberhart.Particle Swarm Optimization. Proc.Of the IEEE Int. Conf. on Neural Networks, pp. 1942-1948,1995. Show Context View Article Full Text: PDF (626KB)

[5] Ciro attaianese, Vito Nardi, Aldo Perfetto, GiuseppeTomasso.Vectorial torque control: a novel approach to torqueand flux control of induction motor drives. IEEE Trans.on Industry Application, vol. 35, pp. 1399-1405, Nov/Dec.1999.Show Context View Article Full Text: PDF (241KB)

[6] W.J. Lukitsch.Soft start vs AC drives Understand thedifferences. Conference Proceedings of IEEE Textile Fiber andFilm Industry, pp. 1-5, 1999.Show Context View Article FullText: PDF (347KB)

[7] ANFIS based soft-starting and speed control of AC voltagecontroller fed induction motor M. Ayyub 2006 IEEE PowerIndia Conference Year: 2006 Page: 5 pp. Cited by: Papers (4)IEEE Conferences

[8] Ant Colony Optimization - a tool for online tuning of a PIcontroller for a three phase induction motor drive JasdeepKour. Sheela Tiwari 2013 International Conference on Control.Automation Robotics and Embedded Systems (CARE)

20


8206

8207

8208

Documents

STARTING CURRENT REGULATION OF INDUCTION MOTOR … · Comparatively star/delta starting is the most preferable ... motor smooth starting is a dynamic process and we ... the rst place