4

Click here to load reader

ANALYSIS AND COMPARISON OF MAXIMUM POWER … · ANALYSIS AND COMPARISON OF MAXIMUM POWER POINT TRACKING ALGORITHMS FOR PV SYSTEMS ... 2.4 FUZZY LOGIC CONTROL ... vast …

Embed Size (px)

Citation preview

Page 1: ANALYSIS AND COMPARISON OF MAXIMUM POWER … · ANALYSIS AND COMPARISON OF MAXIMUM POWER POINT TRACKING ALGORITHMS FOR PV SYSTEMS ... 2.4 FUZZY LOGIC CONTROL ... vast …

ANALYSISANDCOMPARISONOFMAXIMUMPOWERPOINTTRACKINGALGORITHMSFORPVSYSTEMS

BRAHMANSINGHBHALAVI

AssistantProfessor,SEES,DAVV,INDORE,INDIA

ABSTRACT

ThemaximumpowerpointTracker(MPPTs),forthesolarpanelsplaysanimportantrolebecauseitprovidesthemaximumoutputfromthePVsystem,foragivensetofconditionsoftheirelectricpowersystems,andthereforemaximizestheefficiencyof thepanel,which therebyhelps inminimizing the totalsystemcost.PresentlyanumberofMPPTalgorithmsareavailableformaintainoperationat themaxi‐mumpowerpoint,however,everyalgorithmshas theirownadvantagesand limitationswhichcausesthedifferentbehaviorwhenusedincommercialsolarpowerMPPTs.Thispaperisintendedtopublisheffectivecomparison amongst the different algorithms and represent an optimized solution on the basis of requirement specificcriteria.Theresults from thisworkcanbeutilized to find thebestMPPTsystemdependinguponspecificrequirementsandresourcesavailability.

Keywords:MaximumPowerPointTracking(MPPT),SolarPanel,PhotovoltaicSystems(PV).

1. INTRODUCTION

Under uniform illumination, PV array has a constantcurrent voltage (IV) characteristic onwhich there is aunique point on the IV curve known as the maximumpower point (MPP). The array can be operated in thehighest efficiency to produce a maximum output power.WhenthePVarrayisconnectedtotheload(theso‐called"directly coupled" systems) itmaydirect thePVpanel todifferent operation points. In general, depending on theoperating point it is difficult to achieve the MPP. Toovercomethisproblem,MPPTconverterareusedtotrackmaximumpowerpoint,andmaintaintheoperatingpointofthePVarrayattheMPP.

If properly controlled by the algorithm,MPPT can locateandtracktheMPPinPVarray.However,thelocationoftheMPP in the plane is not known in advance, henceitshould be located, or by means of model calculations orthrough a search algorithm. Furthermore the situationisagain complicated by the fact that MPP dependsnonlinearly on the light and temperature, as shown infigure1(a) under increasing radiation at a constanttemperature, and Figure 1 (b) shows the curvesunder the same values of the light, but at a highertemperature. This paper presents a variety of ways fordiscussion of each algorithm. In this paperminormodifications of different existing methods are avoidedanddiscussedundermainmethod.

Figure1(a):SolarcellI‐Vcharacteristicsfordifferentirradiationvalues.

Figure1(b):Solarcellpowercharacteristicsfordifferentirradiationvalues.

Manuscriptconcludeswithadiscussionondifferentwaysdependingontheirimplementation,thenecessarysensors,theirabilitytodetectmany localMaxima, theircosts,andthey suit applications. A summary of the key features ofdifferentmethodsarealsoprovided.

BRAHMAN SINGH BHALAVI et al. DATE OF PUBLICATION: FEB 19, 2015

ISSN: 2348-4098 VOL 3 ISSUE 1 JAN-FEB 2015

INTERNATIONAL JOURNAL OF SCIENCE ENGINEERING AND TECHNOLOGY- www.ijset.in 294

Page 2: ANALYSIS AND COMPARISON OF MAXIMUM POWER … · ANALYSIS AND COMPARISON OF MAXIMUM POWER POINT TRACKING ALGORITHMS FOR PV SYSTEMS ... 2.4 FUZZY LOGIC CONTROL ... vast …

2. MPPTALGORITHMS

As explained earlier, the MPPT algorithm is required inordertogetthemostpowerfromsolarpanels.TheMPPTalgorithms are used to track the MPP of a solar panelwhich largelydependson the radiationand temperature.Over the last decade, many methods are developed andpublishedtofindtheMPPwithreliability.Thesemethodsdiffer in many aspects, such as the necessary sensors,complexity, cost, range, rate of convergence,effectivenessof trackingunder thedynamicradiationandtemperaturevariations. Presently P&O (Perturb and Observe) and In‐Cond(IncrementalConductance)algorithmsare themostcommonly used for MPPT. These methods have theadvantage of a simple implementation, but they alsohavedisadvantages,aswillbeshownlater.Othermethodsbased on different principlesfuzzy logic control, neuralnetworks, open circuit voltage and the fraction of shortcircuit current, the current sweep, etc. most of thesemethods give a local maximum, and some, such asfractional open circuit or short circuit current, giveapproximate and multiple MPP’s which is helpfulspecificallyforpartiallyshadedPVARRAY’s,whereseveralMPP’s can exists. In the next section, some of the mostpopularMPPTtechniquesarediscussed.

2.1HILL‐CLIMBINGTECHNIQUE

P&OandIn‐Condalgorithmsarebasedontheprincipleofthe"Hillclimb",whichconsistsofmovingtheoperationofPV in the direction in which the power increases. Hillclimbing techniques are the most popular methods ofMPPT, thanks to their ease of implementation and goodperformance, forconstantirradiation. The advantages ofboth methods are simplicity and low computationalcomplexity.AlsotheyhavewellknowndisadvantagessuchasvibrationsaroundMPPandcompletelyfailstotracktheMPPduringarapidlychangingoperationalconditions.

2.2PERTURBANDOBSERVE(P&O)

The P&O algorithm is also known as “hill‐climbing”,hencebothnamesrefertothesamealgorithmconceptthe only difference is how it is implemented. The Hill‐climbingmethodinvolvesaperturbationonthedutycycleof theDC toDCconverterandP&Oaperturbation in theoperating voltage of the DC connection between the PVarrayand theDC toDCconverter. In thecaseof theHill‐climbing, perturbation in the duty cycle of the powerconverterisperformedtochangethevoltageoftheDClinkbetween thePVarrayand theDC toDCconverter, hencebothreferstothesameconcept.

Finallythetechniqueutilizespreviousgrowthdisturbancestodecidewhatshouldbe thenextdisturbancedependinguponthechangeinpower.Ifthereisanincreaseinpower,proceduremustbefollowedinsamedirectionotherwiseitmustbemovedintheoppositedirectionandthisprocessisrepeateduntil it reaches theMPP.However inpractice itnever stabilized on a MPP instead it fluctuates aroundMPP.

Figure2:PerturbandObserve(P&O)AlgorithmFlowChart

2.3INCREMENTALCONDUCTANCE

Incremental conductance algorithm utilizes thecharacteristiccurvebetweenpowerandvoltage(current)of PV panelwhich shows the zero slope at MPP andpositiveornegativeontotheotherrespectivesides.

Comparing the change in increment of voltage againstoutput (current)between two consecutive measurementsofthePVpaneltherequiredvoltagechangeforMPPcanbedetermined.

Figure3:IncrementalConductanceAlgorithmFlowChart

InbothP&OandIn‐CondschemesthetimetoachieveMPPdepends on the size of the MPPincrement the referencevoltage. Besides providing quicker convergence thesealgorithms has to main limitations. The first and mostimportant of them, that theycan easily lose track of theMPP if the radiation changes quickly, althoughin case

BRAHMAN SINGH BHALAVI et al. DATE OF PUBLICATION: FEB 19, 2015

ISSN: 2348-4098 VOL 3 ISSUE 1 JAN-FEB 2015

INTERNATIONAL JOURNAL OF SCIENCE ENGINEERING AND TECHNOLOGY- www.ijset.in 295

Page 3: ANALYSIS AND COMPARISON OF MAXIMUM POWER … · ANALYSIS AND COMPARISON OF MAXIMUM POWER POINT TRACKING ALGORITHMS FOR PV SYSTEMS ... 2.4 FUZZY LOGIC CONTROL ... vast …

ofstep changes they trackMPP very well, because of theinstantaneous change whichdoes not requirecontinuouschanging of the curve. However, when environmentalconditionsmodifiesthecurve,onwhichthealgorithmsarebasedon;then changes in voltage and current donotonlybecauseoftheperturbationofvoltage.Asaresult,itisnotpossible foralgorithms to determine whether changes inthePVpanelpowerisduetovoltageperturbationorduetoachangeinexposure.

Another obstacle is both voltage and current fluctuatesaroundMPP in the steady state. This is due to the factthatcontrol is discrete and voltage and current do notconstantly remains at MPP, butoscillates around it. Themagnitudeoftheoscillationsdependsontherateofchangeof the reference voltage. The bigger it is, the higher theamplitudefluctuations. However the frequency ofoscillationisinverselyproportionaltothestepsizeoftheincrement of voltage. The traditional solution is atradeoff between oscillations and tracking time as if theincrement is small so that the oscillations decrease,then the MPP is reached slowly and vice versa, so acompromisesolutionhas tobe foundThus, the last threeMPPTmethodsarebasedonthesameprinciples,P&OandIn‐Condalgorithms,sotheyhavethesameadvantagesanddisadvantages.All Climbing Hill MPPT methods ofPhotovoltaic array depend on V‐P or‐P featuresthatdepends on the temperature and radiation, so thesemethods MPPT can beconfused when radiation ortemperaturechanges.Finally,othermethodsofHillclimbMPPT does not offer any improvement inthe originalalgorithmsP&OandIn‐Cond.

2.4FUZZYLOGICCONTROL

Theuseof fuzzy logiccontrolhasgainingpopularityoverthe traditional control systems because it candeal withcomplexsystemswithoutanaccuratemathematicalmodeland canalso handle the nonlinearity. Recent growth indigital electronicssuch as microprocessors andmicrocontrollers also helped in popularizing fuzzy thecontrollogic.

Figure3:FuzzyLogicControlledMPPTBlockDiagram

InBasicstructuretheFuzzylogiccontrollercanbedividedintothreephases:fuzzy‐fication,inferencesystemsandde‐fuzzy‐fication.

Figure4:fuzzy‐ficationandde‐fuzzy‐ficationmembershipfunctions.

Fuzzy‐fication process extracts linguistic variables basedon degree of membership for certain sets from inputnumeric values.Membership functions, are used toassociate a class membership or relation for eachthelinguistic notion. The number of membership functionsrequiredisdependsontheaccuracyofthecontroller.

Figure5:fuzzyinferencesystemrulebase.

The inference systems is simple a lookup table of ruleswhich used to estimate the output from each linguisticvariablescombinations(ANDorOR).Finallytheoutputofinference systems for all derived linguistic variablescombinations of numeric inputs are combined using de‐fuzzy‐ficationtoproducerequiredcontroloutput.

2.5NEURALNETWORKS

Neural networks came along with Fuzzy Logic and bothare the part of Soft Computing techniques. However theneural network is completely different from fuzzy logic.The neural network or more accurately artificial neuralnetworkisamathematicalmodelofbio‐neuronsdesignedto gain the information from given set of training datasamples. The tracking accuracy of such systems dependuponnumberofneuronsandlayersofneuronsandqualityoftrainingdataset.

BRAHMAN SINGH BHALAVI et al. DATE OF PUBLICATION: FEB 19, 2015

ISSN: 2348-4098 VOL 3 ISSUE 1 JAN-FEB 2015

INTERNATIONAL JOURNAL OF SCIENCE ENGINEERING AND TECHNOLOGY- www.ijset.in 296

Page 4: ANALYSIS AND COMPARISON OF MAXIMUM POWER … · ANALYSIS AND COMPARISON OF MAXIMUM POWER POINT TRACKING ALGORITHMS FOR PV SYSTEMS ... 2.4 FUZZY LOGIC CONTROL ... vast …

F

Theneedeverdifficchanperi

3.

Theprevsectiuseimpl

Thispoinalgoexteresupoinstratof bwithstreacondhighprovcontbutbaseperfcomnotdue

REF

[1]MaxTRAJUNE

[2]PickContEner

[3]CZhenMPP7,21

Figure6:Sim

main complded for thepryPVarrayancult becausenge with tiodicalre‐trai

MAXIMUMSUMMARY

vastmajorityvious yearsions.Someo the same lementedin

sstudyshownt trackingrithmsthatantofthestudults proposent trackingtegy,isthemusiness convhallotherstramlined forductance perher executionvide a bittrollerbasedrequiresmiced techniquformance instmpletecharactpractically fetopanelagin

FERENCES

Trishan Esrximum PowANSACTIONSE2007.

Saleh Elkelakert “Overvietrol MethodsrgyEngineeri

Chun‐LiangLn Yang “AnPTAlgorithm177‐2193.

mple3‐LayerN

lexity in thisprocesshas tndforeverye the charactime, so theining.

POWERANDCONC

yoftheMPPThave beenofthemarefundamentaldistinctivew

wsanexaminaefficienciesareexamineddywasrestrithat, on theeffectivenes

mostgenerallverters, can prategiesconsr the giverforms similn cost wouldbetter perfmethodcoulcrocontrollerue could btantaneouslyteristicsinforeasible alsongcanalsoaff

ram “CompawerPoint TrONENERGY

ani Babaa, Mew of Maxims for PV Systing,2014,2,

Liu,Jing‐HsiaAsymmetricaforPhotovol

NeuralNetwo

s system is ttobespecificoperatingcoteristics of ae neural ne

R POINTCLUSION

Talgorithmsreviewed iextremelyl concepthowways.

ationof themof a few

d inpreviousictedtothosepremise ofms, the pertlyutilizedalgpossibly bemsidering thaten equipmelar to P&O,d not be advformance. Tdprovidemurs, finally thebe used toyhowever formationofththe change iffectitsperfor

rison of Phracking TecCONVERSION

Matthew Armum Powertems”, Journ59‐72.

oChen,Yi‐Hual Fuzzy‐LogltaicSystems

orkStructure

the trainingcallyacquirednditionswhia panel mayetwork requ

T TRACK

createdoverin the prevcomparableweverapplie

maximumpoMPPT con

ssectionsanealgorithms.maximum poturb‐and‐obsgorithmasamost competit is legitimant. Incremeyet becauseocated evenThe Fuzzy luchbetterrese neural netwo achiever that it requhepanelwhicin characterirmance.

otovoltaic Achniques”, IN,VOL.22,N

mstrong, VoPoint Trac

nal of Power

uaLiuandZgic‐Control‐Bs”,Energies2

e.

datad forchisalsouired

KING

rtheviousandedor

owerntroldtheTheowererveparttitiveatelyentale itsif itlogicsultsworkbestuireschisstics

ArrayIEEENO.2,

olkerckingand

ong‐ased014,

[4]“MaTur

[5]AbdNeuInteApp

[6](20EnvSym

[7]M. (PhoIEE262

[8]AT.MSyst201

[9]Revand564

[10]MaxSystSoc

[11]PickConEne

Ali M. Eltaaximum Powrbines”,http:/

Rihab Mahjdallah “Maximural Networkernational Joplication,201

Sera, D., Ker06) ImprovevironmentalmposiumonI

Femia,N.,Lis(2008)DistriotovoltaicArrE Transactio21.

Ali,A.N.A.,SaM. (2012)AStems. 201212,1‐17.

Narendiran,view.2013IndComputingT4‐567.

]Barakati, Mximum PowetemIncludingietyGeneral

] Saleh Elkekert “Overvintrol MethodergyEngineer

maly, A. I.wer Extraction//dx.doi.org/

oub Essefi,mum Powerks for Stand‐ournal of Mo14,3,53‐65.

rekes, T., Teoed MPPT MConditions.ndustrialEle

si,G.,PetronibutedMaximrays:NovelAons on Indu

ied,M.H.,MourveyofMaxIEEE Energy

S. (2013)GrnternationalCTechnologies

M., Kazerani,er TrackinggaMatrixCoMeeting,24,

elaniBabaa,ew of Maxis for PV Sysring,2014,2,

Alolah andn fromUtility/10.5772/546

Mansour SoPoint Tracki‐Alone Photoodern Nonlin

odorescu, R.Method for R

2006 IEEctronics.

e,G.,SpagnumumPowerApproachandustrial Electr

ostafa,M.Z.anximumPPTTytech, Cleve

ridTie InvertConferenceo(ICCPCT),20

M. and AplControl fornverter.IEEE705‐713.

Matthew Armum Powerstems”, Journ59‐72.

Hassan M.y‐InterfacedW675.

ouissi, Hsaning Control Uovoltaic Systenear Theory

and BlaabjerRapidly ChanEE Internat

olo,G.andVPointTrackindSystemAnaronics, 55, 2

ndAbdel‐MonTechniquesoland, 29‐31

terandMPPonCircuits,P0‐21March2

levich, D. (2a Wind TurEPower&En

rmstrong, Vr Point Tracnal of Power

FarhWind

HadjUsingems”,y and

rg, F.ngingtional

Vitelli,ngofalysis.2610‐

neim,ofPVMay

T—Aower2013,

2009)rbinenergy

olkerckingr and

BRAHMAN SINGH BHALAVI et al. DATE OF PUBLICATION: FEB 19, 2015

ISSN: 2348-4098 VOL 3 ISSUE 1 JAN-FEB 2015

INTERNATIONAL JOURNAL OF SCIENCE ENGINEERING AND TECHNOLOGY- www.ijset.in 297