用人工神经网络控制策略跟踪光伏发电系统(IJISA-V6-N12-3)

I.J.IntelligentSystemsandApplications,2014,12,17-26PublishedOnlineNovember2014inMECS()DOI:10.5815/ijisa.2014.12.03Copyright©2014MECSI.J.IntelligentSystemsandApplications,2014,12,17-26TrackingPowerPhotovoltaicSystemusingArtificialNeuralNetworkControlStrategyM.T.MakhloufiLEALab.Electronicsdepartment,FacultyofTechnology,BatnaUniversity,ChahidM.ohamedBelhadiBoukhloufRoad,Batna,AlgeriaE-mail:[e-mail],Y.Abdessemed,A.BoutarfaLRP&LEALabs.Electronicsdepartment,FacultyofTechnology,BatnaUniversity,ChahidM.ohamedBelhadiBoukhloufRoad,Batna,AlgeriaE-mail:[e-mail],[e-mail],[e-mail]—Photovoltaicgenerationisthetechniquewhichusesphotovoltaiccelltoconvertsolarenergytoelectricenergy.Nowadays,PVgenerationisdevelopingincreasinglyfastasarenewableenergysource.However,thedisadvantageisthatPVgenerationisintermittentbecauseitdependsconsiderablyonweatherconditions.Thispaperproposesanintelligentcontrolmethodforthemaximumpowerpointtracking(MPPT)ofaphotovoltaicsystemundervariabletemperatureandsolarirradiationconditions.Inthispaper,asimulationstudyofthemaximumpowerpointtracking(MPPT)foraphotovoltaicsystemusinganartificialneuralnetworkispresented.ThesystemsimulationiselaboratedbycombiningthemodelsestablishedofsolarPVmoduleandaDC/DCBoostconverter.FinallyperformancecomparisonbetweenartificialneuralnetworkcontrollerandPerturbandObservemethodhasbeencarriedoutwhichhasshowntheeffectivenessofartificialneuralnetworkscontrollertodrawmuchenergyandfastresponseagainstchangeinworkingconditions.IndexTerms—SolarEnergy,Photovoltaic,MPPT,P&O,BoostConverter,ArtificialNeuralNetworkI.INTRODUCTIONSignificantprogresshasbeenmadeoverthelastfewyearsintheresearchanddevelopmentofrenewableenergysystemssuchaswind,seawaveandsolarenergysystems.Amongtheseresources,solarenergyisconsiderednowadaysasoneofthemostreliable,dailyavailable,andenvironmentfriendlyrenewableenergysource[1-2].However,solarenergysystemsgenerallysufferfromtheirlowefficienciesandhighcosts[3].Inordertoovercomethesedrawbacks,maximumpowershouldbeextractedfromthePVpanelusingMPPTtechniquestooptimizetheefficiencyofoverallPVsystem.MPPTisareal-timecontrolschemeappliedtothePVpowerconverterinordertoextractthemaximumpowerpossiblefromthePVpanel[12].TheMPPTworkingprincipleisbasedonthemaximumpowertransfertheory.Thepowerdeliveredfromthesourcetotheloadismaximizedwhentheinputresistanceseenbythesourcematchesthesourceresistance.Therefore,inordertotransfermaximumpowerfromthepaneltotheloadtheinternalresistanceofthepanelhastomatchtheresistanceseenbythePVpanel.Forafixedload,theequivalentresistanceseenbythepanelcanbeadjustedbychangingthepowerconverterdutycycle[4].TheliteratureisrichwithvariousMPPTtechniquesbasedondifferenttopologiesandwithvaryingcomplexity,cost,andoverallproducedefficiency[13].TheHillClimbing(HC)andthePerturbandObserve(P&O)arethemostknownandcommerciallyusedtechniques[5-7].OthermodifiedmethodssuchastheincrementalConductance(INC)technique,theneuralnetwork(ANN)technique,andfuzzylogiccontrollertechnique,havebeenalsoreportedtoimprovetheperformanceofthesetechniques.InHC-MPPTtechnique,thedutycycleisdirectlyincrementedordecrementedinfixedstepsdependingonthepanelvoltageandpowervaluesuntilthemaximumpowerpoint(MPP)isreached.TheP&OtechniquesharesthesameHCconceptofoperation,butwithanadditionalPIcontrolloop.IntheP&O,theconverterinputreferencevoltageistheperturbedvariableandthedutycycleiscomputedthroughanadditionalPIcontrolloop[16].TheadditionalcontrolloopresultsinanincreaseintheP&Oefficiency,asthesystemdemonstratesafasterdynamicperformanceandbetter-regulatedPVoutputvoltagecomparedtoHC.TheP&Omethodiscommonlyusedbecauseofitssimplicityandeaseofimplementation[5-6].Furthermore,P&O(withasmallstepsize)innominalconditionscanhaveMPPTefficienciesmostlythesamelikeothercomplextechniques,andstilleasierimplementation.However,thedrawbackofthistechniqueisthattheoperatingpointofthePVarrayoscillatesaroundtheMPP.Therefore,thepowerlossmayincrease.Furthermore,whenthesuninsolationchangesrapidly,theP&OmethodprobablyfailstotracktheMPP.AnotherpossibledisadvantageisthattheMPPTmaynotbeabletolocatetheMPPastheamountofsunlightdecreases,becausethePVcurveflattensout[5].RecentlyintelligentbasedcontrolschemesMPPThavebeenintroduced.18TrackingPowerPhotovoltaicSystemusingArtificialNeuralNetworkControlStrategyCopyright©2014MECSI.J.IntelligentSystemsandApplications,2014,12,17-26Inthispaper,anintelligentcontroltechniqueusingartificialneuralnetworkcontrolisassociatedtoanMPPTcontrollerinordertoimproveenergyconversionefficiency.Thesimulationcangeneratetwodifferentsolutionsforthecontrolofconvertersystem;oneisP&OcontrollerandtheotheroneisANNcontroller.ThecircuitdiagramoftheenergyconversionsystemisshowninFig.1.Thesystemconsistsofphotovoltaicpanel,aDC/DCboostconverter,acontrolunitandaresistiveload.Thefirststageofthesystemissolarpanel.TheI-Vcharacteristicofapaneldependsonthetemperatureandsolarirradiance.ThethreemostimportantcharacteristicsofPVpanelaretheshortcircuitcurrent,opencirc