Seite - 161 - in Emerging Technologies for Electric and Hybrid Vehicles

Energies 2017,10, 5 methodis that theparameterextractiontest corresponds toaspecificoperatingscenario. If theactual loadprofilesshowobviouslydifferentbandwidthsunderdifferentworkingconditions, theparameter extractiontest shouldbere-implemented.Onesolutiontoovercomethisdrawbackis toconductas manyparameterextractiontestsaspossible tocover the typical loadcharacteristics,but this requires anextensiveamountof timeandeffort. 1.2. ContributionsofThisPaper Basedonthebatteryparameterestimationmethodsdiscussedabove, it canbeconcludedthat seldomdoeswork in theprevious literaturediscussabatterymodelconsideringboth theCCcharging anddynamicdriving scenarios. Hence, the focus of this paper is to propose a batteryparameter estimationmethod,which is applicable to commonoperating scenarios inHEV/EVapplications. Themaincontributionsare: (1)both theconstant-currentchargingandthedynamicdrivingscenarios aretakenintoconsideration,andtwoseparatesetsofmodelparametersareestimatedthroughdifferent partsof thepulse-rest test; (2) themodelparameters for theconstant-current chargingscenarioare estimated from thedata in thepulse-chargingperiods; (3) themodel parameters for thedynamic drivingscenarioareestimatedfromthedata in therestperiods,andthe lengthof thefitteddataset isdeterminedbythespectrumanalysisof the loadcurrent; (4) theunsaturatedphenomenoncaused by the long-termRCnetwork is analyzed, and the initial voltage expressions of theRCnetworks in thefitting functionsare improved toensureahighermodelfidelity; (5)both thesimulationand experimentresultsagreewith theanalysisanddemonstrate the improvementof theproposedbattery parameterestimationmethodover theexistingones. 2. ParameterExtractionProcedure 2.1. ParameterExtractionTestDesign It can be seen fromFigure 1 that the second order ECMcontains oneOCV-SoC relationship and five impedance parameters (Rin,Rshort,Cshort,Rlong andClong), which need to be estimated. Theoretically,allof the impedanceparametersmentionedaboveshouldbemultivariable functionsof SoC, theC-rateof the loadcurrent (C is theamplitudeof thecurrentwithwhich thebatterycanbe fullydischargedin1h), temperatureandcyclenumbers [39,45]. These functionsnotonlymakethe parameterextractionprocesscomplexandtimeconsuming,butalsoincreasethecomputationalburden of theBMS.Hence,withincertainerror tolerance, somerelationships canbesimplifiedor ignored. Usually, aging periods are generally in the range ofmonths to years. While for the system-level simulations of automotive applications, the time periods of interest are typically in the range of seconds tohoursordays inspecial cases [43,45].Hence, the long-termagingeffect isusually ignored in theparameterestimationprocessandhandledseparately inmostcases [39,46]. In this paper, all of themodel parameters are estimated through the discharging/charging pulse-rest test at room temperature (22 ◦C–25 ◦C). A lithium-ion polymer battery with nickel- manganese-cobalt-basedcathodeandgraphite-basedanode isunder test. Its specificationsaregiven inTable1,andthedetailedexperimental stepsaredescribedas follows. Table1.Specificationof the testedbattery. ChargeCapacity 40.99Ah Dischargecapacity 40.89Ah Nominalvoltage 3.7V Chargecutoffvoltage 4.2V Dischargecutoffvoltage 2.7V Thedischargingpulse-rest test startswitha fully-chargedbattery. Ineachcycleof the test, the battery is discharged at a 2%SoC stepwithC/2 constant current, then followedby a rest period. 161