电子测量技术ELECTRONICMEASUREMENTTECHNOLOGY第45卷第23期2022年12月DOI:10.19651/j.cnki.emt.2209936基于遗忘因子递推最小二乘法的锂电池等效电路模型参数辨识方法*赵可沦江境宏邓进刘洪飞(广州广电计量检测股份有限公司广州510656)摘要:本文对现有常用的锂离子电池模型进行分析,建立了便于工程应用的二阶RC网络等效电路模型,并在MATLAB中搭建相应电池模型,利用实测数据对电池模型参数进行离线辨识,对模型精度进行了验证。考虑到模型参数值在电池充放电过程中并不恒定,而是受到充放电倍率和电池荷电状态等因素影响不断变化,因此为提高模型精度,采用含遗忘因子的递推最小二乘方法进行模型参数的在线辨识,通过仿真分析对比不同遗忘因子的影响,确定了遗忘因子的最佳范围。实验结果表明,随着遗忘因子从1开始减小,模型的精度会先提高再减小。本模型比较合适的遗忘因子范围大致为0.90~0.95,最佳值应在0.94附近,此时模型的平均电压误差仅为0.00043V,证明了本文辨识方法的正确性和高精度。关键词:电池模型;参数辨识;最小二乘法;遗忘因子中图分类号:TM912.9文献标识码:A国家标准学科分类代码:470ParameteridentificationmethodoflithiumbatteryequivalentcircuitmodelbasedonforgettingfactorrecursiveleastsquaresZhaoKelunJiangJinghongDengJinLiuHongfei(GuangzhouGRGMetrologyandTestCo.,Ltd.,Guangzhou510656,China)Abstract:Inthispaper,theexistingcommonlyusedlithium-ionbatterymodelsareanalyzed,andasecond-orderRCnetworkequivalentcircuitmodelwhichisconvenientforengineeringapplicationisestablished.ThecorrespondingbatterymodelsarebuiltinMATLAB,andthemeasureddataareusedtoidentifythebatterymodelparametersoffline,andtheaccuracyofthemodelisverified.Consideringthatthevaluesofmodelparametersarenotconstantintheprocessofbatterycharginganddischarging,butconstantlychangeduetofactorssuchascharginganddischargingrateandbatterySOC,inordertoimprovetheaccuracyofthemodel,therecursiveleastsquaremethodwithforgettingfactorsisadoptedtoidentifythemodelparametersonline,andtheoptimalrangeofforgettingfactorsisdeterminedbycomparingtheinfluencesofdifferentforgettingfactorsthroughsimulationanalysis.Theexperimentalresultsshowthatastheforgettingfactordecreasesfrom1,theaccuracyofthemodelwillfirstincrea...
