Page - 160 - in Emerging Technologies for Electric and Hybrid Vehicles

Energies 2017,10, 5 5VKRUW5VKRUW 5ORQJ5ORQJ 2&92&9 5LQ5LQ &ORQJ&ORQJ&VKRUW&VKRUW 7LPH V 7LPH V 9ROWDJH9ROWDJH &XUUHQW&XUUHQW 9W9W ,, 9 9 9 9 , , , , $$ %% Figure1.Thesecondorderequivalentcircuitmodel (ECM).OCV, opencircuitvoltage. Generally speaking, batteries usually operate in two scenarios in automotive applications: The constant-current (CC) charging scenario and thedynamic driving scenario [27]. Usually, the motions of lithium ions under the continuous external excitation (representing the CC charging scenario)andthediscontinuousexternalexcitation(representingthedynamicdrivingscenario) show different characteristics, and this difference is related to the diffusivity of ions. In otherwords, themodelparameters, especially theRCnetworkparameters, showdiversevaluesunderdifferent operatingscenarios [21,28]. Therefore,batteryparametersshouldbe identifiedseparatelyaccordingto theactualoperatingscenarios.Abundant researchworkhasbeenconductedtoseek theaccurateECM for thespecificoperatingscenario. For thechargingscenario, auniversalmodelbasedonasimple mathematicalequationwithconstantparameters isproposed[29–31]. Themathematicalequations includeonepolynomialcomponentandoneor twoexponential functions,andrelevantparameters canbeobtainedbyfittingcollectedchargingprofiles.Verificationresults inrelated literatureshowthat theoverallmodeloutputprofilesmatchwellwith theexperimentaldata,but therestill existsobvious estimationerrorsduringcertainperiods (at thebeginningof theplateauregionandthe last charging region). This ismainlycausedbytheconstantparametersduringthewholechargingprocesssince theactualmodelparameters, suchas timeconstants,mayvarygreatlyatdifferentSoC regions [32]. Theworks in [32–34] estimate themodel parameters through the data in the rest periods of the pulse-rest testatdifferentSoCpoints, andtheestimatedmodelparameterscanbeshownas functions ofSoC.However, thechargingconcentrationprocessundercontinuousexcitation isdifferent fromthe chargingrecoveryprocessunder therestperiod[19,35]; thus, theestimatedmodelparametersmaynot accuratelyrepresent thechargingcharacteristicsof thebattery. For thedynamicdrivingscenario,many modelingapproacheshavebeenreportedonthebasisofthepulsedischargeanalysis. In[36–38],model parametersareobtainedbysimplealgebraicoperations. This is straightforward,but largeestimation errorsexist. Amoreaccuratemethodis tofit thevoltageresponseof thewhole restperiodwithan exponential function[39–41]. The limitationof thismethodis itspoordynamicperformance. Inorder to improvethebatterymodelaccuracy,HuandWangin[42]proposea twotime-scale identification algorithmtoseparate the identificationsofslowandfastbatterydynamics. Thismethodshowsbetter frequencyresponsematchingwithout increasingcomputationalcomplexity. Xiongin[17]uses thebias correctionmethodtoensure thebatterymodelpredictionperformance. Thisapproachshowsexcellent performanceandhighaccuracyagainstuncertainoperatingscenariosandbatterypacks. Insteadof the conventionalpulse-rest test, [43,44]propose twotypesofapplication-orientedparameterextraction tests, leadingtoa fastdynamicsbatterymodelwithhighfidelity.Onemajor limitationof thiskindof 160