Comparison of Minimising Total Harmonic Distortion with PI Controller, Fuzzy Logic Controller, BFO- fuzzy Logic Controlled Dynamic Voltage Restorer

Transcription

1 Interntionl Journl of Eletroni nd Eletril Engineering. ISSN , Volume 7, Numer 3 (24), pp Interntionl Reserh Pulition House omprison of Minimising Totl Hrmoni Distortion with PI ontroller, Fuzzy Logi ontroller, FO- fuzzy Logi ontrolled Dynmi Voltge Restorer Shilp Grg nd Rm vtr Jswl 2 M Teh Student (EE), UIET, Kurukshetr University, Kurukshetr, Indi Deprtment of EE, UIET, Kurukshetr University, Kurukshetr, Indi strt This pper desries the minimistion of totl hrmoni distortions in distriution line. DVR is used for this purpose. ut the ontrolling of DVR voltge injetion is done y three different ontrollers: PI, Fuzzy Logi nd FO-Fuzzy Logi. omprison is shown mongst these three tehniques. Every tehnique is heked for voltge sg nd swell. Keywords: voltge sg, voltge swell,dynmi voltge restorer(dvr), PI ontroller, Fuzzy ontroller, FO-Fuzzy logi,totl hrmoni distortion.. Introdution Power qulity is n issue tht is eoming inresingly importnt to eletriity onsumers t ll levels of usge. With the inrese in the use of sensitive equipments nd non liner lods in the domesti s well s industril re, n enhned wreness of power qulity is developed mongst eletriity onsumers. However presene of non liner lods nd non zero impedne of the supply system my generte lrge numer of inequlities nd disturnes in the system whih ultimtely leds to poor power qulity.[,3] These disturnes rete situtions in whih the wveform of the supply voltge (voltge qulity) devite from the sinusoidl wveform for ll three phses of three-phse struture t rted frequeny with mplitude orresponding to the rted Root Men Squre (RMS) vlue. numer of power qulity disturnes my enountered in the system. The dequte rnge of power qulity disturnes overs sudden, short durtion vritions viz. impulsive nd osilltory trnsients, voltge sgs, short interruptions, s well s stedy stte devitions, suh s hrmonis nd fliker. These prolems re tkled y series onneted devie nmed Dynmi Voltge Restorer (DVR). It is sed on pulse width modulted Voltge soure inverter,whih is ple of generting or soring rel or retive power independently.[3 6]

2 3 Shilp Grg & Rm vtr Jswl The DVR hs two modes of opertion whih re, the stndy mode nd the oost mode. In the stndy mode (VDVR=), the ooster trnsformer s low voltge winding is shorted through the inverter. No swithing of semiondutors tkes ple in this mode of opertion. Therefore, only the omprtively low ondution losses of the semi-ondutors in this urrent loop ontriute to the losses. In the oost mode (VDVR>), the DVR will injet ompenstion voltge through the ooster trnsformer due to detetion of supply voltge disturne. 2. Proposed lgorithm DVR is used to ontrol injeted voltge into distruution line to ompenste fults whether sg or swell. For this purpose IGT ontrolled y pulse width modultor is used. The pulse width of PWM n e ontrolled nd for this purpose some ontrollers like PI, fuzzy logi nd FO optimised fuzzy logi, re used here. Fuzzy logi lgorithm s performne depends upon its memership funtions nd rule sets. When rules nd memership funtions re designed for speifi tsk then these re onsidered to e fixed. ut due to hnge in system onditions stedy stte error hnges. So to minimise stedy stte errors, memership funtion should e ontinuously hnging with initil ondition hnge. So position of memership funtions is optimised. In our work this is done y teril forging optimistion (FO). FO is disussed lter in this pper. Time onsumed in itertions to minimise the stedy stte error n e set y itertion numer nd numer of teri s onsidered. Totl numer of positions in memership funtions, deides serhing spe dimensions. In this work s shown in Fig. 2., 7 memership funtions for eh input nd output re used. Positions of these re fixed initilly. Two types of memership funtions re used: trpezoidl nd tringulr. In error input, either one or two position vetor of funtion oinides, this redues the dimension of serhing spe to gret extent. Tringulr funtion hs three positions [x, x2, x3], stisfying ondition x<x2<x3. s is seen in Fig. 2., the middle position of NM oinides with strting position of NS nd similr other memership funtions lso hve some ommon positions, reduing totl of 4 positions to e optimised to minimise the stedy stte error for single input. Fig. 2.: Memership funtion of input error

3 omprison of Minimising Totl Hrmoni Distortion with PI ontroller, Fuzzy 3 Sine there re two inputs nd one output, so totl of 2 positions of memership funtions re to e tuned y FO lgorithm. It mens in FO every single teri will serh in twelve serhing spes for nutrition. During optimistion n ojetive funtion is deided whih minimise the stedy stte error. For every teri twelve positions of memership funtions re hnged nd new fuzzy logi is evluted for given error. Here in this work integrl squre error (ISE) is tken to ojetive funtion nd used s input for evlution of fuzzy logi. fter ll omplete itertions error is minimised nd tht instntneous positions of memership funtions re used s new positions in fuzzy logi. 3. teril Forging Optimistion teri Forging Optimiztion lgorithm (FO), is new omer to the fmily of nture-inspired optimiztion lgorithms. For over the lst five dedes, optimiztion lgorithms like Geneti lgorithms (Gs), Evolutionry Progrmming (EP), whih drw their inspirtion from evolution nd nturl genetis, hve een dominting the relm of optimiztion lgorithms. Reently nturl swrm inspired lgorithms like Prtile Swrm Optimiztion (PSO), nt olony Optimiztion (O) hve found their wy into this domin nd proved their effetiveness.. teri serh for nutrients in mnner to mximize energy otined per unit time.. terium tkes forging deisions fter onsidering two previous ftors. The proess, in whih terium moves y tking smll steps while serhing for nutrients, is lled hemotxis.during forging of the rel teri, loomotion is hieved y set of tensile flgell. When they rotte the flgell in the lokwise diretion, eh flgellum pulls on the ell. Tht results in the moving of flgell independently nd finlly the terium tumles with lesser numer of tumling wheres in hrmful ple it tumles frequently to find nutrient grdient.when they get food in suffiient, they re inresed in length nd in presene of suitle temperture they rek in the middle to from n ext repli of itself. This phenomenon inspired Pssino to introdue n event of reprodution in FO. Due to the ourrene of sudden hnges, the hemotti progress my e destroyed nd group of teri my move to some other ples or some other my e introdued in the swrm of onern. This onstitutes the event of elimintiondispersl in the rel teril popultion, where ll the teri in region re killed or group is dispersed into new prt of the environment.[2,4] Thus serh for food of E.oli n e tegorised into four steps: hemotti, Swrming, Reprodution nd Killing/ Dispersion. Mthemtilly these n e represented step y step s: hemotti: θ (j +, k, l) = θ (i) (j, k, l) + (i) (i) (i) Where θ (j, k, l) represents i-th terium t jth hemotti, k-th reprodutive nd l-th elimintion-dispersl step. (i) is the size of the step tken in the rndom diretion speified y the tumle (run length unit). Swrming J (i, j, k, l) = J (i, j, k, l)+ J ( θ( j, k, l),p( j, k, l)

4 32 Shilp Grg & Rm vtr Jswl where J (i, j, k, l) is the fitness funtion.[2, 4] 4. Results & Disussion efore optimising fuzzy logi, rules nd memership funtion hve to e defined. In our se dvr.fis nmed fuzzy logi struture is designed in MTL whose deision tle is shown in tle. Tle : Fuzzy Rule tle[5] E\E N NM NS Z PS PM P N N N N NM NM NS Z NM N N NM NM NS Z PS NS N NM NM NS Z PS PM Z NM NM NS Z PS PM PM PS NM NS Z PS PM PM P PM S Z PS PM PM P P P Z PS PM PM P P P Memership funtion defined in dvr.fis re shown in Fig. 4. elow. Fig. 4.: Memership funtion () error signl () Error rte () tuting Output Signl.

5 omprison of Minimising Totl Hrmoni Distortion with PI ontroller, Fuzzy 33 Fuzzy logi funtion tkes two inputs: error nd error rte nd gives one output whose memership funtions re shown in Fig. 4.. In this pper totl hrmoni distortions re minimised nd ompred using PI, fuzzy logi nd FO trined fuzzy logi for voltge sg nd swell. The proposed simulink model is shown in ppendix. Fourier trnsform is pplied on output signl to hek THD (%). The output wveform nd fft nlysis of output in se of PI ontroller is seleted y mking onstnt vlue 3 for swith seletion s shown in ppendix. voltge sg is pplied y fult reker within intervl.-.3. Votge sg in time intervl Sg Removl y PI ontroller Fig. 4.2(): Output of PI ontrolled sg. Fundmentl (5Hz) =.6582, THD=.46% Mg (%of Fundmentl) Frequeny (Hz) Fig. 4.2(): THD fter FFT nlysis of sg introdued output.. Fundmentl (5Hz) =.9888, THD=.24% Mg (% of Fundmentl) Frequeny (Hz) Fig. 4.2(): THD fter FFT nlysis of sg ompensted output

6 34 Shilp Grg & Rm vtr Jswl Similrly the ompenstion of voltge sg is done y fuzzy logi nd fo-fuzzy logi nd THD y FFT nlysis is noted down. The hnge in memership funtion fter FO optimistion is shown in Fig New positions of eh memership funtion is otined fter optimistion stisfying ondition for tringulr memership funtion s disussed ove. tle showing the omprison of THD(%) for every se disussed is shown in tle 2. Fig. 4.3: Optimised Memership funtion () error signl () Error rte () tuting Output Signl. Tle 2: omprison of THD THD (%) Without ontrol.46 PI.24 Fuzzy Logi.2 FO- Fuzzy Logi.8 The ompenstion of voltge swell y proposed lgorithm is shown in Fig. 4.4 elow.

7 omprison of Minimising Totl Hrmoni Distortion with PI ontroller, Fuzzy 35 Votge Swell in time intervl Swell Removl y FO-Fuzzy ontroller Fig. 4.4: Voltge Swell ompenstion. 5. onlusion The ove shown results show tht fuzzy logi minimistion of THD proves etter thn PI ontroller. Performne of fuzzy system is more inresed if memership funtions re optimised y ny optimistion lgorithm. In this pper teril forging optimistion is used s triner to memership funtions. Tle shown in results depits lerly tht fter FO optimistion THD is minimum s ompred to others. Work desried in this pper works well for oth sg nd swell ompenstion. Referenes [] M.Shrny,.svrj, M.Ssikl, n Overview of Dynmi Voltge Restorer for Voltge Profile Improvement Interntionl Journl of Engineering nd dvned Tehnology (IJET) ISSN: , Volume- 2, Issue-2, Deemer 22 [2] Mohmmd KINI, Seyed Mohmmd li MOHMMDI, teril forging optimiztion pproh for tuning type-2 fuzzy logi ontroller Turkish Journl of Eletril Engineering & omputer Sienes Turk J Ele Eng & omp Si (23) 2: [3] Rosli Omr, N.. Rhim nd Mrizn Sulimn, Dynmi Voltge Restorer pplition for Power Qulity Improvement in Eletril Distriution System: n Overview ustrlin Journl of si nd pplied Sienes, 5(2): , 2 [4] Sushree Sngit Ptnik nd nup Kumr Pnd, Prtile Swrm Optimiztion nd teril Forging Optimiztion Tehniques for Optiml urrent Hrmoni Mitigtion y Employing tive Power Filter Hindwi Pulishing orportion pplied omputtionl Intelligene nd Soft omputing Volume 22, rtile ID [5].Rjni, Dr.P.Sngmeswr Rju, omprision Of PI, Fuzzy & Neuro- Fuzzy ontroller sed Multi onverter Unified Power Qulity onditioner IJEET Volume 4, Issue 2, Mrh pril (23), pp. 36-5

8 36 Shilp Grg & Rm vtr Jswl [6] Rosli Omr, N.. Rhim nd Mrizn Sulimn, Dynmi Voltge Restorer pplition for Power Qulity Improvement in Eletril Distriution System: n Overview ustrlin Journl of si nd pplied Sienes, 5(2): , 2 ppendix Disrete, Ts =. s. po we rgui reker Sg reker2 Three-Phse Soure Three-Phse Trnsformer (Three Windings) T3 Out Voltge Swell Sg Sope O/P Voltge Line Feeder Line Feeder 2 FO Voltge Swell Doule lik on this to run optimistion 3 onstnt =Fuzzy ontrol 2=FO -Fuzzy ontrol 3. PI ontrolled ontrol_seleti on Goto Note:. For Sg--- reker is open nd reker 2 is losed 2.For Swell--- reker 2 is open nd reker is losed In Fig. 5: Proposed Model for omprison of PI, Fuzzy Logi nd proposed ontroller for THD redution. DVR T V Three-Phse reker g + - Universl ridge V T3 Sope2 Mg Phse In Pulse s Uref Disrete PWM Genertor Sope4 v inv_re dlt Susystem ontrol_seletion From MT L Funtion Fuzzy Logi MTL Funtion FO Fuzzy Logi Disrete PI ontroller 3-Phse Sutrt Sequene nlyzer du/dt onstnt Derivtive u 2 input s Mth Integrtor To Workspe Funtion PI Multiport Swith Fig. 6: DVR Model for proposed work hving swith for PI, fuzzy Logi nd FO fuzzy logi