文档名：基于多时间尺度双扩展卡尔曼滤波的电池峰值功率估计方法
摘要：动力电池是电动汽车的技术瓶颈,其状态的高精度估计一直是行业的技术难点,不准确的状态估计值易造成安全隐患,并加速动力电池系统老化.然而,动力电池每用必衰、时变非线性、环境敏感性等特点导致对其状态的实时精准估计极具挑战性.该文针对锂离子动力电池峰值功率估计的问题,提出基于多时间尺度滑动窗口的双扩展卡尔曼滤波(DEKF)算法,基于峰值功率测试结果更新模型参数库,实现了参数的缓时变估计.评价指标显示,动力电池全寿命、全电量区间内,变温度等条件下的验证结果表明所提算法能够准确估计电池参数和功率状态,电压误差小于40mV.
Abstract：StateofPower(SOP)estimationisoneofthecorefunctionsofbatterymanagementsystem(BMS).CurrentSOPestimationmethodsaremainlydividedintothreecategories:characteristicmapmethod,data-drivenmethodandmulti-constraintmodel-basedestimationmethod.However,inpracticalapplications,traditionalSOPestimationmethodsaregenerallydifficulttoobtainaccurateandrealisticpeakpower.Therefore,alargenumberofbatterycell,moduleandbatterypackexperimentsarecarriedoutunderdifferenttemperatures,differentworkingconditionsanddifferentagingstatesinthispaper.Abatterycellmodelandbatterypackmodelwithdual-characteristiccellsuniformlydistributedareestablishedtoimplementtheexpansionofthecellmodeltothesystemalgorithm,whichishighpredictionaccuracyandlowcalculationamount.Inaddition,adoubleKalmanfilter(DEKF)algorithmbasedonmulti-timescaleslidingwindowsisproposed.Themodelparameterlibraryisupdatedbasedonthepeakpowertestresults,whichrealizestheslowtime-varyingestimationoftheparametersandimprovestheaccuracyandrobustnessofthealgorithmtoestimatethepeakpower.Firstly,anequivalentcircuitmodel(ECM)isestablishedasthebatterymodelinthispaper.Then,bycomparing12commonECMsintermsofmodelingaccuracyandcomputationalcomplexity,theTheveninmodelisfinallyselectedtosimulatebatterycharacteristics,duetotherelativelyhighvoltagepredictionaccuracyandrelativelysmallamountofcalculation.Basedonthebatterycellmodel,abatterypackmodelwithdualcharacteristiccellsuniformlydistributedisestablished.Thetwocharacteristiccellsarethereal-timehighestcellvoltageandthelowestcellvoltageinthebatterypack.Thecellnumbersofcharacteristiccellsarehencechangeable.Atthesametime,theparallellinkisregardedasalargecell,andthemodelparameterschangewiththebatterycharacteristics.Basedontheassumptionofuniformdistributionofbatterycells,thehighestandlowestcellvoltageandbatterypackvoltageareusedasinput.Thecalculationformulaofthebatterysystemissimplified,andthetransformationfromthesinglemodeltothesystemmodelisrealized.FortheTheveninmodel,theopen-circuitvoltageattimek+1istakenasthefirst-orderTaylorexpansionattimek,andthepowerbatteryvoltagecalculationmodelisobtained.Consideringthatthepeakpowerhasapositivecorrelationwiththeterminalvoltage,thedichotomymethodisusedtotrytoselectacertainpeakpower.Bycomparingthecorrespondingterminalvoltagewiththetargetterminalvoltage,thedichotomymethodisusedtocorrectthepeakvaluePowerlevel,toachievethepurposeofconstantlyapproachingtherealpeakpower.Finally,ahardware-in-the-loop(HIL)simulationplatformwasbuilt.Theexperimentalobjectsmainlyincluded50A?hthree-cellbatterycellsand12-stringbatterymodules.Onthebasisoftheinitialcharacteristictestofpowerbatterycellsandmodules,representativecellsandmodulesthatneedtobetestedareclassifiedandscreened,andthenahalf-yearfull-lifeandmulti-temperaturepowerpredictioniscompletedinaccordancewiththeexperimentalrequirementsandproceduresverificationandagingcycletestexperimentsareconducted,andalargeamountofexperimentaldataisobtained.TheresultsshowthattheimprovedDEKFalgorithmiseffective,hashighaccuracyandrobustness,andtheSOCestimationerrorislessthan3%.Effectivevoltageerrorislessthan40mV.
作者：李强 张凯旋 袁文文 许亚涵 杨瑞鑫 方煜 Author：LiQiang ZhangKaixuan YuanWenwen XuYahan YangRuixin FangYu
作者单位：潍柴动力股份有限公司潍坊261061北京理工大学机械与车辆学院北京100081
刊名：电工技术学报
Journal：TransactionsofChinaElectrotechnicalSociety
年，卷(期)：2024, 39(7)
分类号：TM912
关键词：电动汽车 锂离子电池 峰值功率估计 双扩展卡尔曼滤波算法 滑动窗口
Keywords：Electricvehicle lithiumionbattery peakpowerestimation doubleKalmanfilter(DEKF)algorithm slidingwindow
机标分类号：U469.72TM912.9TP311.13
在线出版日期：2024年4月12日
基金项目：国家自然科学基金基于多时间尺度双扩展卡尔曼滤波的电池峰值功率估计方法[
期刊论文] 电工技术学报--2024, 39(7)李强 张凯旋 袁文文 许亚涵 杨瑞鑫 方煜动力电池是电动汽车的技术瓶颈,其状态的高精度估计一直是行业的技术难点,不准确的状态估计值易造成安全隐患,并加速动力电池系统老化.然而,动力电池每用必衰、时变非线性、环境敏感性等特点导致对其状态的实时精准估计极...参考文献和引证文献
参考文献
引证文献
本文读者也读过
相似文献
相关博文
基于多时间尺度双扩展卡尔曼滤波的电池峰值功率估计方法 Lithium-Ion Battery Peak Power Estimation Based on Multi-Time Scale Double Extended Kalman Filter
基于多时间尺度双扩展卡尔曼滤波的电池峰值功率估计方法.pdf
- 文件大小:
- 1.42 MB
- 下载次数:
- 60
-
高速下载
|
