Pulse Density Modulation Pattern Optimization Using Genetic Algorithms (D. Pimentel IECON2006)

Preview:

Citation preview

Pulse Density Modulation Pattern Pulse Density Modulation Pattern Optimization using Genetic Optimization using Genetic

AlgorithmsAlgorithmsDemian Pimentel, Ahmed Chériti,

Mohamed Ben Slima and Pierre Sicard

Presented at IECON 2006Presented at IECON 2006

By Pierre Sicard, Member, IEEEBy Pierre Sicard, Member, IEEE

November 2006November 2006

Prepared by D. PimentelPrepared by D. Pimentel

22

PresentatiPresentation layouton layout

33

I. Power electronics devicesII. Pulse-density modulation (PDM)III. Genetic algorithms (GA)IV. Application of GA to PDM pattern

generationV. Simulation resultsVI. Conclusions

Presentation layoutPresentation layout

44

I.Power Power electronielectroni

cs cs devicesdevices

55I.Power electronics Power electronics devicesdevices

Undesirable effects of PE devices on power distribution lines

66

Desired characteristics of new PE designs

Few total harmonic distortion (THD)High power factor (PF)High conversion efficiencyEntirely digital controlsCost and size reduction

I.Power electronics Power electronics devicesdevices

77

II.PDMPDM

88

Modulation technique used for resonant converters

Advantages compared to PWM▲ Higher power factor▼ Lower total harmonic distortion

▶Lower electromagnetic noise ◀Ease of implementation

DisadvantagesPF decreases as pulse-density decreasesTHD increases as pulse-density decreasesDiscrete output characteristicNon-linear power vs pulse-density curve

II.PDMPDM

99II.PDMPDM

Experimental Power Factor and THD for a PDM one-phase converter using 16 modulation levels

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

6 13 19 25 31 38 44 50 57 63 69 75 81 88 94 100

Pulse density (%)

Power factor

THD (%)

1010II.PDMPDM

0

10

20

30

40

50

60

70

80

90

100O

utp

ut

Po

we

r (%

)

6 13 19 25 31 38 44 50 57 63 69 75 81 88 94 100

Pulse density (%)

analyticalapproximation

simulation

2max seqoutput NseqPP

Output power vs. pulse density for a PDM one-phase converter using 16 modulation levels

1111II.PDMPDMSwitching modes

1212

Non-regular PDMpattern

generation

II.PDMPDM

Logic Control signals

1313

III.Genetic Genetic algorithmalgorithm

ss

1414

Rarely used for PE applicationsCan be employed for

OptimizationSearch problems

JargonChromosomesPopulationGenerationEnvironmentFitness function

GoalBring individuals within a population to

“evolve” towards desired solutionsStochastic process

III.Genetic algorithmsGenetic algorithms

1515III.Genetic algorithmsGenetic algorithmsAG representation and flow chart

1616III.Genetic algorithmsGenetic algorithmsAG common operators

1717

IV.GA GA applicatiapplicati

onon

1818

PDM optimal pattern generation using GAGA algorithm programmed using Matlab®

Simulink® was used to evaluate fitness▶Simulation of power system (SimPowerSystems)

Research spacePDM sequences are represented by bit stringsSynchronization of PDM sequences with line voltage

▶Length equals 1/120 s▶Resonant frequency = 15.96 kHz▶String length = 133 resonant cycles

Mutation operator was used“1” represents a pulse“0” represents a dead pulse

IV.GA applicationGA application

1919IV.GA applicationGA application

Simulink® block diagram

2020

Relationship betwen PF and THD

Fitness function

Termination conditions

IV.GA applicationGA application

21

cos

THDPF

targetedevaluatedTHD

targetedevaluatedPF

THDTHDfitness

PFPFfitness

stop

0

and

0

THD

PF

fitness

fitness

Pulse Densi

ty

8/ 133

16/ 133

24/ 133

32/ 133

40/ 133

PF 0.60.75

0.80.85

0.9

THD 1.3 0.8 0.7 0.6 0.5

2121

V.SimulatiSimulation on

resultsresults

2222V.Simulation resultsSimulation results

Load voltage produced by several AG-generated patterns with different pulse densities

2323V.Simulation resultsSimulation resultsLine current created by several AG-generated patterns with different pulse

densities

2424V.Simulation resultsSimulation results

Two similar AG-generated patterns

2525V.Simulation resultsSimulation results

Comparison: AG patterns perform significantly better.

2626

VI.ConclusioConclusionsns

2727

Significant improvement of PFSignificant reduction of THDEasy practical implementation of GA

patterns (single microcontroller)GA algorithm could be used to improve PF in

the range of 25-95% pulse-densityAlgorithm could be hardware-implemented

for auto-tuning devicesComponent agingChanging noisy environmentsLoad variationsSwitching frequency change

Additional parameters optimisationDC-link capacitor

VI.ConclusionsConclusions

2828

VII.ReferencReferenceses

2929

K. Bose, “Energy, Environment, and Advances in Power Electronics,” IEEE Transactions on Power Elec-tronics, Vol. 15, No. 4, pp. 688-701, 2000.

M. M. Morcos and J. C. Gómez, “Electric Power Qual-ity, The Strong Connection with Power Electronics,” IEEE power and energy magazine, pp. 18-25, Septem-ber / October 2003.

H. Fujita and H. Akagi, "Control and Performance of a Pulse-Density-Modulated Series-Resonant Inverter for Corona Discharge Process,” IEEE Transactions on Industry Applications, vol. 35, no. 3, pp. 621-627, May / June 1999.

H. Fujita and H. Akagi, "Pulse-Density-Modulated Power Control of a 4 kW, 450 kHz Voltage-Source In-verter for Induction Melting Applications,” IEEE Transactions on Industry Applications, vol. 32, no. 2, pp. 279-286, March / April 1996.

J. Essadaoui, P. Sicard, É. Ngandui and A. Chériti, “Power Inverter Control for Induction Heating by Pulse Density Modulation with Improved Power Factor,” IEEE Canadian Conference on Electrical and Com-puter Engineering – Toward a Caring and Human Technology, Canada, pp. 515-520, May 2003.

A. Sandali, A. Chériti, and P. Sicard, “Comparison of the Various PDM Control Modes,” IEEE International Conference on Industrial Technology, pp. 574-579, 2004.

Fairchild Semiconductor, “Induction Heating System Topology Review,” Discrete Application, Power De-vice Division, Fairchild Semiconductor, AN9012, June 2000.

J. M. Corrêa, E. D. Hutto, F. A. Farrett, and M. G. Simões, “A Fuzzy-Controlled Pulse Density Modula-tion Strategy for a Series Resonant Inverter with Wide Load Range,” Power Electronics Specialists Confer-ence, vol.4, p. 1650-1655, June 2003.

M.P. Kazmierkowski, M Cichowlas, and M. Jasinski, “Artificial Intelligence based controllers for industrial PWM power converters,” IEEE International Confer-ence on Industrial Informatics, p. 187-191, August 2003.

B. Ozpineci, J. O. P. Pinto, and L. M. Tolbert, “Pulse-Width Optimization in a Pulse Density Modulated High Frequency AC-AC Converter Using Genetic Algo-rithms,” IEEE International Conference on Systems, Man, and Cybernetics, vol. 3, p. 1924-1929, 2001.

VII.ReferencesReferences

Recommended