A Genetic-based Fuzzy Logic Power System Stabilizer

8/8/2019 A Genetic-based Fuzzy Logic Power System Stabilizer

1/6

A GENETIC-BASED FUZZY LOGIC POWER SYSTEM STABILIZER FORMULTIMACHINE POWER SYSTEMS

M. A. Abido Y. L. Abdel-MagidElectrical Engineering Departm entKing Fahd University of Petroleum and MineralsDhahran 31261, Saudi ArabiaAbstract - This paper presents a novel approach tocombine genetic algorithms (GA) with fuzzy logicsystems to design a genetic-based fuzzy logic powersystem stabilizer (GFLPSS) or multimachine powersystems. Incorporation of G A in fuzzy logic powersystem stabilizers (FLPSSs) design will significantlyreduce the time consumed in the design process ofFLPSSs. t is shown in this paper that the performanceof FLPSS can be improved significantly byincorporating a genetic-based learning mech anism. Theperformance of the proposed GFLPSS under differentdisturbances is investigated. The results show the

superiority of the proposed GFLPSS as compared to theclassical PSS and its capability to enhance systemdamping to local as well as interarea modes ofoscillations. The capability of the proposed GFLPSS owork cooperatively with the existing classical PSSs salso demonstrated.1. INTRODUCTION

Due to increasing complexity of electrical powersystems, there has been increasing interest in thestabilization of such systems. In the past two decades,the utilization of supplementary excitation controlsignals for improving the dynamic stability of powersystems and da mping out the low frequency oscillationsdue to disturbances has received much attention [l-91.Nowadays, the conventional power system stabilizer(CPSS) a fixed parameters lead-lag compensator - iswidely used by power system utilities [3]. The gainsettings of th ese stab ilizers are determined based on thelinearized model of the power system around a nominaloperating point to provide optimal performance at thispoint. Generally, the power systems are highly nonlinearand the operating conditions can vary over a wide range.Therefore, CPSS performance is degraded whenever theoperating point changes from one to another because offixed parameters of the stabilizer.

Alternative controllers using adaptive controlalgorithms have been proposed to overcome suchproblems [5-61. However, most adaptive controllers aredesigned on the ba sis of parameter identification of thesystem model in real-time which results in timeconsuming and com putational burden.

Recently, fuzzy logic power system stabilizers(FLPSSs) have been proposed 17-91. FLPSSs appear tobe the most suitable stabilizers due to their lowercomputation burden and robustness. Unlike the mostclassical methods, an explicit mathem atical model of thesystem is not required to design a good FLPSS whichmakes it more suitable for on-line computer control. Inaddition, FLPSS can be easily set up usingmicrocomputer with A/D nd D/A onverters [101.Although fuzzy logrc controllers showed promisingresults, they are subjective and somewhat heuristic. Inaddition, generation of membership functions, and thechoice of scaling factors are done either iteratively, bytrial-and-error, or by hum an experts. There is to-date nogeneralized method for the formulation of fuzzy controlstrategies, and design remains an ad hoc trial and errorexercise. That makes the design of fuzzy logic controllera laborious and time-consuming task.On the other hand, genetic algorithms (GA) re:search algorithms based on the mechanics of naturdselection and survival-of-the-fittest notion. Theprinciples of GA were first introduced by Holland in hspioneering work in the theoretical development andadaptation in natural and amficial systems [ll].Recently, GA have been applied to various power systemproblems with promising results [12-141.

The recent approach is to integrate the use of GAand fuzzy logic systems in order to combine theirdifferent strengths [15-16]. A genetic-based fuzzy logicpower system stabilizer for a single machine infinite bussystem has been proposed with promising results [171.In this paper, we extend this approach to amultimachine power system where the p roblem becomesmore realistic and complex. The results show that theperformance of FLPSSs can be significantly improvedby incorporating a genetic-based learning mechanism.2. GENETIC ALGORITHMS

GA are exploratory search and optimizationprocedures that were devised on the principles of naturalevolution and population genetics. Unlike otheroptimization techniques, GA work w ith a population ofindividuals represented by b it s trings and m o w thepopulation with random search and competition. The

0-7803-4053-1/97/$10.00@ 1997 IEEE 329

Authorized licensed use limited to: Sona College of Technology. Downloaded on August 02,2010 at 05:34:49 UTC from IEEE Xplore. Restrictions apply.


2/6

advantages of GA over other traditional optimizationtechniques can be sum marized as follows:0 GA work on a coding of the parameters to beoptimized, rather than the param eters themselves.0 GA searc h the problem space using a p opulation oftrials rep resenting possible solutions to th e problem,not a sin gle point, i.e. GA have im plicit parallelism.This property ensure s GA to be less susceptible to

getting trapped on local minima.0 GA use a performance index assessment to guidethe search in the problem space.0 GA use proba bilistic rules to make de cisions.Typically, the GA starts with little or no knowledgeof the correct solution dependmg entirely on responsesfrom interacting environment and their evolutionoperators to arrive at optimal or near optimal solutions.In general, GA in clude operations such as reproduction,crossover, and mutation. Reproduction is a process inwhich a new generation of population is formed byselecting the fittest individ uals in the current population.Crossover is the most dominant operator in GA. It isresponsible for producing new offsprings by selectingtwo strings and exchanging portions of their structures.The new offsprings may replace the weaker individualsin the population. Mutation is a local operator w hc h isapplied with a very low probability. Its function is toalter the value of a random position in a string.

@ ( x ; a , b , c )=

3. FUZZY LOGIC CONTROL SCHEME [17]

'0.0 V x I(9)[----12- a v x 1 a , b ]

1 - 2[---]*- c v x E 3 b , c [c - a

c - a1.0 v x 2 c

The stabilizing signal u is added to the excitationloop as shown in Fig. 1.At time t, u(t ) s given byu(t)= U(k) ;kT,< t < ( k+ l )Ts (1)

The value of U(k) is determined at each sampling timebased on fuzzy ogic throug h the following steps:Step 1: Consider the jth machine where the stabilizerwill be installed.Step 2: The speed deviation of the jth machine, do.@),is measured at every sampling time, and theacceleration of the mac hine,AJ{k), s calculated

bY

Step 3: Compute the scaled acceleration,A,fk), using

where Fa j is the scaling factor for the jthmachine acceleration.Step 4 : The generator condition is given by the pointG(k)where

Step 5: CalculateR,(k) and eI(k)using

Step 6: Compute the values of membership functionsN,{ej) an d Pg{0j) defined as [7]

(10)(x;c- , ~b 12 , ~ )XIl-@(~;c,c+b/2,c+b)X>CY(x;b,c)=

It is worth mentioning that these continuousnonlinear membership functions are moresuitable for power system stability problem [7].Step 7; Determine the value of the gain function G,(k)defined as [XI

Step 8: Compute the stabilizing signal v ( k ) using

where Umarj s the maximum value of thestabilizing signal at the t h m achine.Step 9: Consider another machine where the stabilizershould be installed and goback to step 2.Step 10 : Increase k by 1 and return to step 1.

330



3/6

The main tuning parameters of an FLPSS are B v ,Faj , and D j . For the optimal settings of theseparameters, a quadratic performance index J isconsidered:

Vr

N LJ = E [ k T , A o i ( k ) l 2

Externalb

i = l k = l

where N is the number of machines and L is the totalnumber of data po ints.In the above index, the speed deviation of the ithmachme Awi(k) is weighted by the respective time kT,.The index J is selected because it reflects small settlingtime, small steady state error, and small overshoots. Thetuning parameters are adjusted so as to minimize theindex J.

V I /-

DIAICROCOMFUTER AID A@

Fig. 1: Study system configuration.4. THE PROPOSED APPROACH

The major problem in the design of conventionalFLPSSs is that the stab ilizers are designed one at a timeand the parameters of each stabilizer are optimizediteratively [8]. This process is time consuming and theinteraction between stabilizers is not taken into account.This results in degradation of the stabilizer performance.In the proposed approach, all stabilizers are designedtogether and all parameters are optimizedsimultaneously using GA as a n optim ization process.Applying the GA to the problem of PSS designinvolves performing the following two steps.1. The performance index value m ust be calculated foreach of the strings in the c urrent population. To dothis, the tuning parameters must be decoded fromeach string in the population and the system issimulated to obtain the performance index value.GA operations are applied to produce the nextgeneration of the strings.

These two steps are repeated from generation togeneration until the population converges producing anoptimal or near optimal parameter set. Thecomputational flow chart of the design process of theproposed GFLPSS s shown in Fig. 2.The tuning parameters are coded in a b ina q string.The initial popu lation is generated randomly. Population

2.

size, maximum number of generations, and crossoverand mutation probabilities are chosen to be 30, 50, 0.75,and 0.005 respectively. Fig. 3 shows the convergencerate of the performance index J.

initialization

time domain simulation of the systemand performance index evaluation+GA operators

Fig. 2: Flow chart of the proposed GFLPSS esign

i 1.6E-241.4E-2 , I I , I '

0.W 10.M 20.M 30.00 40.00 5 0 . 0GenerationsFig. 3: Variation of the performance index J .5. SIMULATION RESULTS

In this study, the 10-machine 39-bus New E nglandpower system shown in Fig. 4 was considered [8]. Eachmach ine has been represented by a fourth order two-axisnonlinear model. Generator G1 is an equivalent powersource representing parts of the U.S.-Canadianinterconnection system. Details of the system data aregiven in [8]. In this study, the following disturbances areconsidered for the simulations:(a) Three phase fault for 0.10s at bus 29 at the end(b) Three phase fault for 0.15s at bus 15 at the endof line 26-29.of line 14-15.

33



4/6

Fig. 4: Single line diagram for New E ngland system

A0 9

0.021

-0 2

h-3

v)c0ma,Ua,a, -a"

av

.-I.-, .00n ---..............

----

A09

I-

I I I I I I I

0.00 1 oo 2.00 3.00 4.00 5.00Time (s)Fig. 5: S ystem response without PSSs for disturbance (a)0.01 7

h3avmc0mare.-I.-, .00Ua,"

-0.01 ' I l I l I l [ l0.00 1 .oo 2.00 3.00 4.00 5.00Time (s)Fig. 6: System response without PSSs for disturbance (b)

It is worth pointing out that disturbance (a) isconsidered to excite local mode of oscillation of the

generator G9 while disturbance (b) is considered toexcite interarea mode of oscillations of the system.Without PSSs, the system responses due todlsturbances (a) and (b) are shown in Figs. 5 and 6respectively. It is observed that the system damping isvery poor and the system is highly oscillatory.Therefore, it is necessary to install stab ilizers in order tohave good dynamic performance. To i d e n a theoptimum locations of PSSs, the participation factormethod [18] and the sensitivity of PSS effect (SPE)method [19] were used. The results of both methodsindlcate that the generators G5, G7, and G9 are theoptimum locations for installing PSSs to damp out theelectromechanical modes of oscillations. Therefore,these generators are equipped with three of the proposedGFLPSSs.The system responses with the proposed GFLPSSsfor disturbances (a) and (b) are shown in Figs. 7 and 8respectively. In these Figs., the system performancewith the proposed GFLPSSs is compared with thosewith CPSSs and FLPSSs [SI. The results illustrate thesuperiority of the proposed GFLPSSs to CPSSs andFLPSSs and its efficiency to damp out the local modesas well as the inte rarea modes of oscillations.

0'011 h I\

V-0.01Wa,"4

332



5/6

0'01 1 h0m ---W

.-cAm5....... ....... "O6; 0.01

Q,40.02I,.00 2.00 4.00 6.00 8.00Time (s)(c)Fig. 7 : System response for disturbance (a)(a) with CPSSs

(b) with FLPSSs(c) with proposed GFLPSSs6. COORDINATION BETWEEN GFLPSS ANDCPSS

In most situations, the newly installed GFLPSSswill have to work together with CPSSs which alreadyexist in a power system. In th is section, system responsewith the proposed GFLPSSs and CPSSs workingtogether has been also investigated. Severalcombinations between the proposed GFLPSSs andCPSSs are considered as shown in Figs. 9 and 10whereC and P refer to CPSS and Proposed GFLPSSrespectively. The first, second, and third letters in eachcombination denote the type of stabilizer installed onG5 , G7, and G9 respectively.The system responses to the dsturbances (a) and(b) with different combinations are show n in Figs. 9 and10 respectively. It can be seen that the two types of PSSscan work coo peratively. The response with the proposedGFLPSSs and CPSSs combinations is better and theoscillations are damped out much quicker than theresponse with only CPSSs. Generally, the systemperformance is improved as the num ber of the proposedGFLPSSs installed increases as show n in these Figures.

0.01 7 A 9--.............. 6w6

-0.01 I I I I 1a.oo 2.00 4.00 6.00 8.00Time (s)(a)

-3avv)c0G(UW13Ua,Wnv)

.-> 0.00

-0.01 j 10.00 2.00 4.00 6.00 8.00Time (s)

(b)

-0.01..00 2.00 4.00 6.00 8.00Time (s)

(c)Fig. 8: System response for disturbance (b)(a) with CPSSs(c) with proposed GFLPSSs(b) with FLPSSS

7. CONCLUSIONIn this study, a genetic-based fuzzy logic powersystem stabilizer is introduced. The proposed GFLPSSwas designed by incorporating CA to search for theoptimal settings of FLPSS tuning parameters. Thesimulation results show that the performance of FLPSScan be improved significantly by incorporating mgenetic-based learning mechanism. It is shown that theproposed GFLPSS can provide good dampingcharacteristics during transient conditions for local aswell as interarea modes of oscillations. In addition, thecoordination between the proposed GFLPSS and the

conventional stabilizers is demonstrated.8. ACKNOWLEDGMENT

The authors would like to acknowledgethe supportof King Fah d University of Petroleum & Minerals.

333



6/6

----_CCP Combtnabon--- PCP Comb inatm- PP COmblMtIOn-0 2 0 00 2 00 4 00 6 00 8 00Time (5)Fig. 9: System response with different combinations ofPSSs for dsturbance (a)

-0.01

CCC Combinatin

I I I I I I I

___- -PC CombinationI

0.00 2 00 4 00 600 8 00Time (s)Fig. 10: System response with different combinations ofPSSs for disturbance (b)9. REFERENCES

113 P. M. Anderson and A. A. Fouad, Power SystemControl and Stability, Iowa State Univ. Press,Ames, Iowa, U.S.A., 1977.[2] Y. N. Yu, Electric Power System Dynamics,Academ ic Press, 1983.[3] E. Larsen and D. Swann, Applying power systemstabilizers, IEEE Trans. PAS, vol. 100, no. 6,[4] F. P. De Mello and T. F. Laskowski, Concepts ofpower system dynam ic stability, IEEE Trans. PAS,[5 ] Y. Hsu and K. Liou, Design of self-tuning PIDpower system stabilizers for synchronousgenerators, IEEE Trans. EC, vol. 2, no. 3, 1987,[6] D. Xia and G. T. Heydt, Self-tuning controller forgenerator excitation control, IEEE Trans. PAS,

1981, pp. 3017-3046.

vol. 94, 1979, pp. 827-833.

pp. 343-348.

vol. 102, 1983, pp. 1877-1885.

[7] M. Hassan, 0.P. Malik, and G. S.Hope, A fuzzylogic based stabilizer for a synchronous machine,IEEE Trans. EC, vol. 6, no. 3, 1991, pp. 407-413.[SI T. Hiyama and T. Sameshma, Fuzzy LogicControl Scheme for On-Line Stabilization ofmultimachme Power System, Fuzzy Sets andSystems, vol. 39, pp. 181-194, 1991.[9] H. A. Toliyat, J. Sadeh, and R. Ghazi, Design of

augmented fuzzy logic power system stabilizers toenhance power system stability, IEEE Trans. EC,vol. 11, no. 1, 1996, pp. 97-103.[10]K. A . El-Metwally, G. C. Hancock, and 0. P.Malik, Implementation of a fuzzy logic PSS usinga micro-controller and experimental test results, IEEE Trans. EC , vol. 1 1 , no. 1, 1996, pp. 91-96.[111 J. H. H olland, Adaptation in natural and artificialsystems, Addison-Wesley, 1975.[12]R. Dimeo and K. Y. Lee, Boiler-turbine controlsystem design using a genetic algorithm, IEEETrans. EC, vol. 10, no. 4, 1995, pp. 752-759.[13]Y. L. Abdel-Magid and M . M . Dawoud, Geneticalgorithms applications in load frequency control,Conference of Genetic Algorithms in EngineeringSystems: Innovations and Applications, Sept. 1995,[14]P. Ju, E. Handschm, and F. Reyer, Geneticalgorithm aided controller design with applicationto SVC, IEE Proc. Gen. Tran. Dist., vol. 143, no.[15]C.C. Karr and E.J. Gentry, Fuzzy Control of pHUsing Genetic Algorithms, IEEE Trans. on FuzzySystems, vol. 1, no. 1, pp. 46-53, 1993.[16]D. Park, A. Kandel, and G. Langholz, Genetic-Based New Fuzzy Reasonoing Models withApplication to fuzzy Control, IEEE Trans. syst.,Man, Cybern., vol. 24, no. 1, pp. 39-47, 1994.[17]M. A. Abido and Y. L. Abdel-Magid, Tuning of afuzzy logic power system stabilizer using geneticalgorithms,Accepted fo r Presentation in 4th IEEEInternational Conference on EvolutionaryComputation E E C 97, Indianapolis, USA, April[lS]Y. Y. Hsu and C. L. Chen, Ide ntification ofoptimum location for stabilizer applications usingparticipation factors, IEE Proc., pt. C, vol. 134,no. 3, May 1987, pp . 238-244.[19]E. 2. Zhou, 0. P. Malik, and G. S. Hope, Theoryand m ethod for selection of power system stabilizer

location, IEEE Trans. on Energy Conversion, vol.6, no. 1, 1991,pp. 170-176.

pp , 207-213.

3, 1996, pp. 258-262.

13-16, 1997.

334

Date post:	10-Apr-2018
Category:	Documents
Upload:	dwarakesh14
View:	224 times
Download:	0 times

A Genetic-based Fuzzy Logic Power System Stabilizer

Documents