VOLTAGE SAG IMPROVEMENT BY PARTICLE SWARM OPTIMIZATION OF FUZZY LOGIC RULE BASE

Transcription

1 VOL., NO. 7, PRIL 206 ISSN RPN Journl of Engineering nd pplied Sienes sin Reserh Pulishing Network (RPN). ll rights reserved. VOLTGE SG IMPROVEMENT Y PRTILE SWRM OPTIMIZTION OF FUZZY LOGI RULE SE sl Ni nd N. lert Singh 2 Deprtment of Eletril nd Eletronis Engineering, Noorul Islm University Thukly, Tmilndu, Indi 2 SNL Exeutive SNL Ngeroil, Tmilndu, Indi E-Mil: slni@gmil.om STRT In this pper improvement in voltge sg y using PSO optimized fuzzy ontroller is desried. Dsttom is the FTS devie used in voltge sg improvement. Prtile swrm optimiztion (PSO) is used to optimize the if then rules of the fuzzy ontroller. In this system Dsttom is pled in three phse system to ontrol the voltge sg. fuzzy ontroller is designed to ontrol the output of Dsttom. The whole system is simulted using MTL Simulink. The fuzzy ontrolled Dsttom output is ompred with PI ontrolled Dsttom output. The system without Dsttom is lso simulted using MTL Simulink. The fuzzy ontroller rules re optimized using prtile swrm optimiztion nd the results re lso ompred with other systems. Keywords: voltge sg, dsttom, fuzzy logi, optimiztion, fuzzy logi ontroller, prtile wrm optimiztion.. INTRODUTION Power qulity is set of eletril oundries tht llows n equipment to perform in its speified mnner. Voltge sg nd voltge swell re the ommon power qulity issues. Voltge sg is defined s derese in voltge to etween 0. nd pu in rms voltge for durtion of 0.5 yle to minute [2]. Voltge sg ours minly due to fults nd the short iruit urrent results in derese of voltge [6]. Grphil representtion of voltge sg is shown in Figure-. In this pper we onsider voltge sg due to three types of fults nmely single line to ground fult (SLG), doule line to ground fult (DLG), nd three phse fult. s the demnd of eletri power is inresing dy y dy, the trnsmission networks re found to e very wek so tht they nnot supply unrelile supply with good qulity. Flexile trnsmission systems (FTS) is n ide developed sed on power eletroni ontrollers, whih ontrols the vlues of different eletril prmeters. FTS tehnology mkes use of high speed thyristors for swithing in or out trnsmission line omponents for the required performne of the system. There re different types of FTS devies nmely shunt onneted devies nd series onneted devies []. Dsttom is shunt onneted FTS devie. It is retive soure tht n e ontrolled nd it is ple of soring or generting retive power. Dsttom onsists of oupling trnsformer, voltge soure onverter, D energy storge devie nd neessry ontrol iruits [8]. The dsttom n provide ompenstion in oth indutive nd pitive mode. The V-I hrteristi of dsttom is shown in figure. y using dsttom we n ontrol the effet of voltge sg lso. Figure-. Grphil representtion of voltge sg. Figure-2. lok digrm of Dsttom. Intelligent ontrol tehniques provide method of pproximte resoning tht is similr to humn deision mking proess. They hve fst response time nd high rnge of operting onditions. Fuzzy logi provides forml ide for presenting nd implementing humn knowledge out how to ontrol system [3]. The fuzzy logi ontroller is shown elow. Fuzzifition prt onverts risp input vlues into fuzzy vlues. The knowledge se onsists of dtse 4353

2 VOL., NO. 7, PRIL 206 ISSN RPN Journl of Engineering nd pplied Sienes sin Reserh Pulishing Network (RPN). ll rights reserved. of the plnt. It gives ll the required definitions for the fuzzifition proess.rule se represents the ontrolling system of the network. It is represented s set of if-then rules. Inferene pplies fuzzy reson to rule se to otin the output. Defuzzifition proess onverts fuzzy output to risp vlues. Figure-3. Fuzzy logi ontroller. 2. DSTTOM ONTROL SHEME Dsttom is ontrolled y mens of PWM genertor. The test system output is ompred in n error detetor nd its output is fed to fuzzy logi ontroller. The ontroller will tke neessry ontrol tions nd the ontroller will output n ngle δ whih is phse modulted y the following equtions [5]. V =Sin (ωt+δ-2 /3) (2) V =Sin (ωt+δ+ /3) (3) The phse modulted signls re fed to PWM genertor nd the pulses from PWM genertor will ontrol the opertion of DSTTOM during voltge sg. 3. DESIGN of FUZZY LOGI ONTROLLER The fuzzy logi ontroller designed here onsists of two inputs, nmely error nd hnge in error, nd n output. Seven linguisti vriles re seleted. The linguisti vriles re NE, NEM, NES, ZE, POS, POM nd PO. Tringulr shped memership funtions re hosen for the inputs nd output. rule se with strength of forty nine rules is reted y interonneting different vriles. Fuzzy logi ontroller is implemented y using fuzzy logi toolox of Mtl Simulink. The rule se for inputs nd output using different linguisti vriles re given in the tle elow: V =Sin (ωt+δ) () de e Tle-. Fuzzy logi rulese. NE NEM NES ZE POS POM PO NE NE NE NE NE NEM NES ZE NEM NE NE NE NEM NES ZE POS NES NE NE NEM NES ZE POS POM ZE NE NEM NES ZE POS POM PO POS NEM NES ZE POS POM PO PO POM NES ZE POS POM PO PO PO PO ZE POS POM PO PO PO PO Figure-4. Memership funtion for input. Figure-5. Memership funtion for input

3 VOL., NO. 7, PRIL 206 ISSN RPN Journl of Engineering nd pplied Sienes sin Reserh Pulishing Network (RPN). ll rights reserved. ) Initiliztion eh prtile in the popultion, nd tke X(i) nd V(i) rndomly ) Evlute the ojetive funtion of X(i) nd lulte the vlue of fitness(i) ) Initilize Pest (i) with opy of X(i) d) From the vlues of fitness (i) selet est vlue nd keep it s the new fest Figure-6. Memership funtion for output. 4. OPTIMIZTION USING PSO Prtile swrm optimiztion (PSO) is one of the optimiztion tehniques whih n e used with fuzzy logi ontroller in order to optimize the performne of fuzzy logi ontroller. Here we optimize the fuzzy rules of fuzzy logi ontroller to get the optimized results. In PSO the initil popultion of the system is seleted in rndom mnner nd rehes the optiml solutions y updting the different genertions. In PSO, the potentil solutions, lled prtiles, move through the prolem spe y following the reent optimum prtiles [9]. Every prtile monitors its oordintes in the prolem spe, whih re relted with the est fitness vlue it hs rehed so fr. The est fitness vlue is lso stored nd tht vlue is lled pest. Similrly nother est vlue tht is tken y the prtile swrm optimizer is the est vlue, otined so fr y ny prtile in the neighours of the prtile. This position is lled fest. When prtile tkes ll the popultion s its djent neighours, the est vlue is glol est nd is lled gest. The geneti lgorithm proess n e explined y the following steps e) hoose the prtile with the est fitness vlue from ll the prtiles in the popultion s the gest f) For eh prtile lulte veloity of the Prtile nd updte the vlue of prtile position g) hek the seleted gest vlue is right or wrong h) Repet the proess from step 2 until mximum itertion is rehed The new optimized rule se of fuzzy logi ontroller is given in Figure TEST SYSTEM Test system onsists of 250kv soure whih is onneted to the input of three phse three winding trnsformer. The output terminls of three phse tree winding trnsformer re fed to two rnhes of kv eh. In one of the rnhes Dsttom is onneted for the ompenstion of voltge sg is onneted. In the seond feeder we onneted liner nd vrying lods. Different fults re introdued into this rnh nd the test results re nlysed. de e Tle-2. Optimized fuzzy logi rule se. NE NEM NES ZE POS POM PO NE NE NE NE NEM NES NES NES NEM NE NEM NEM NES NES NES NES NES NES NES NES NES NES NEM NEM ZE NES NES ZE ZE ZE ZE ZE PO POS POS POS POS POS POS POS POM POS POS PO POM POM POM PO PO POS POS POM POM PO PO PO 6. SIMULTION RESULTS The system is simulted using Mtl Simulink Sim Power Systems toolox. three phse fult is pplied for time rnge 0.85s-s with fult resistne of 0.66Ω. The system is simulted for s. In the first se no Dsttom is pled for the ompenstion of voltge sg. In the seond se Dsttom with PI ontroller is introdued in the system. In the third se fuzzy ontrolled Dsttom is operted during the time of fult. In the lst stge prtile swrm 4355

4 VOL., NO. 7, PRIL 206 ISSN RPN Journl of Engineering nd pplied Sienes sin Reserh Pulishing Network (RPN). ll rights reserved. optimized fuzzy logi ontroller is provided. susystem lok is lso provided in simulink model to generte the PWM ontrol signls. Mtl Simulink test model nd the simultion results re shown elow. During fult in the system without dsttom voltge sg ours nd there is no ontrol over the system. When PI ontrolled dsttom is pled in the system voltge sg is mitigted. Fuzzy ontrolled dsttom provides muh more improvement in voltge sg mitigtion. The optimiztion of fuzzy ontroller gve muh more improved result. From the vrious figures given elow we n nlyse the performne of different ontrollers. Voltte pu Disrete, = 5e-005 powergui D Voltge Soure Sope reker Series RL rnh reker Soure + - g Universl ridge PulsesUref Disrete PWM Genertor Series RL rnh Vntrol Sope6 Susystem V Trnsformer (Three Windings) Sope5 delt I V-I Mesurement4 Sope4 z Fuzzy Logi Unit Dely ontroller reker z Unit Dely Figure-7. Simulink model of the test system. Figure-8. Output of the system without Dsttom in the system. V I V-I Mesurement Mg Phse Disrete 3-phse Sequene nlyzer du/dt Derivtive Sope3 Termintor Sope onstnt Termintor reker Fult Mg Phse Disrete 3-phse Sequene nlyzer Series RL rnh2 Series RL rnh Sope2 Voltge pu Voltge pu Voltge pu Figure-9. Output of the system with PI ontrolled Dsttom in the system Figure-0. Output of the system with fuzzy logi ontrolled Dsttom Figure-. Output of the system with prtile swrm optimized fuzzy logi ontrolled Dsttom. 7. ONLUSIONS From the test results we n see the performne of fuzzy ontrolled DSTTOM is etter thn the performne of PI ontrolled Dsttom. The mgnitude of 4356

5 VOL., NO. 7, PRIL 206 ISSN RPN Journl of Engineering nd pplied Sienes sin Reserh Pulishing Network (RPN). ll rights reserved. voltge output of Fuzzy ontrolled Dsttom is higher thn tht of PI ontrolled Dsttom nd it is further inresed y PSO optimiztion. PSO optimized fuzzy ontrolled Dsttom n perform well in voltge sg mitigtion. omintion of different intelligent ontrol tehniques n e pplied for the ontrol of Dsttom. The performne of DSTTOM n e further improved y the pplition of hyrid optimiztion tehniques. REFERENES [] Rjesh Gupt, rindm Ghosh nd vinsh Joshi. 20. Performne omprison of VS sed Shunt nd Series ompenstors Used for Lod Voltge ontrol in Distriution Systems. IEEE Trnstions on Power Delivery. 26(). [9] F. Vldez nd P. Melin Prllel Evolutionry omputing using luster for Mthemtil Funtion Optimiztion. Nfips. Sn Diego, US. pp [0] K. R. Suj, I. Jo Rglend Geneti lgorithm-neuro-fuzzy ontroller (GNF) sed UPQ ontroller for ompensting PQ Prolem. Europen Journl of Sientifi Reserh ISSN X, 78(2): [2] Yn Zhng nd Jovi V. Milnovi Glol Voltge Sg Mitigtion with FTS-sed Devies. IEEE Trnstions on Power Delivery. 25(4). [3] S. Mishr, P. K. Dsh, P. K. Hot, nd M. Tripthy Genetilly Optimized Neuro-Fuzzy IPF for Dmping Modl Osilltions of Power System. IEEE Trnstions on Power Systems. 7(4). [4] M. Vishnuvrdhn nd Dr. P. Sngmeswrrju Neuro-Fuzzy ontroller nd is Voltge Genertor ided UPQ for Power Qulity Mintenne. Interntionl Journl of omputer Theory nd Engineering. 4(). [5] htthry Sourh pplitions of DSTTOM Using MTL/Simultion in Power System. Reserh Journl of Reent Sienes. (IS- 20): [6] Jovi V. Milnovi, Fellow, IEEE nd Yn Zhng Modeling of FTS Devies for Voltge Sg Mitigtion Studies in Lrge Power Systems. IEEE Trnstions on Power Delivery. 25(4). [7] F. Vldez, P. Melin, O. stillo. 20. n improved evolutionry method with fuzzy logi for omining Prtile Swrm Optimiztion nd Geneti lgorithms. ppl. Soft omput. (2): p.p [8] him Singh, P. Jyprksh, Sunil Kumr nd D. P. Kothri. 20. Implementtion of Neurl-Network- ontrolled Three-Leg VS nd Trnsformer s Four-Wire DSTTOM. IEEE Trnstions on Industry pplitions. 47(4). 4357