遗忘因子递推最小二乘法辨识锂离子电池参数  

Identification of Li-ion battery parameters by forgetting factor recursive least square method

作　　者：赵转 曹以龙 杜君莉 史书怀 ZHAO Zhuan;CAO Yi-long;DU Jun-li;SHI Shu-huai(School of Electric Power Engineering,Zhengzhou Electric Power College,Zhengzhou,Henan 450000,China;College of Electronics and Information Engineering,Shanghai University of Electric Power,Shanghai 200438,China;State Grid Henan Electric Power Research Institute,Zhengzhou,Henan 450000,China)

机构地区：[1]郑州电力高等专科学校电力工程学院,河南郑州450000 [2]上海电力大学电子与信息工程学院,上海200438 [3]国网河南省电力公司电力科学研究院,河南郑州450000

出　　处：《电池》2023年第6期629-633,共5页Battery Bimonthly

基　　金：中国博士后科学基金第3批特别资助(站前)(2021TQ0097);河南省高等学校重点科研项目(23B470006)。

摘　　要：递推最小二乘法是辨识锂离子电池等效电路模型参数常见的方法,然而随着递推过程中数据的增加,新数据的生成会受到旧数据的影响,导致误差较大。为此,对锂离子电池的二阶RC等效电路模型进行建模和分析,提出一种用于等效电路模型参数在线辨识的遗忘因子递推最小二乘(FFRLS)法。在动力应力测试实验的基础上,在线辨识等效电路模型参数,利用识别的电路参数对电池电压进行在线预测。通过对比不同遗忘因子(λ)下的端电压均方根误差,发现λ=0.86~0.94为最佳范围。所提算法的精度优于递推最小二乘(RLS)法,验证了算法的可行性和有效性。The recursive least square method was a common method to identify parameters of the equivalent circuit model of Li-ion battery.However,with the increase of data in the process of recursion,the generation of new data would be affected by the old data,resulting in large errors.Therefore,the second-order RC equivalent circuit model of Li-ion battery was modeled and analyzed,a forgetting factor recursive least squares(FFRLS)method was proposed for online identification of equivalent circuit model parameters.On the basis of dynamic stress test experiment,the parameters of equivalent circuit model were identified online,the voltage of battery was predicted online by the identified circuit parameters.By comparing the root mean square error of terminal voltage under different forgetting factors(λ),it was found thatλ=0.86-0.94 was the best range.The accuracy of the proposed algorithm was superior to the recursive least squares(RLS)method,which verified the feasibility and effectiveness of algorithm.

关 键 词：锂离子电池 等效电池模型 递推最小二乘(RLS)法 遗忘因子递推最小二乘(FFRLS)法 参数辨识 

分 类 号：TM912.9[电气工程—电力电子与电力传动]