基于GLD电池模型和FFRLS-EKF算法的SOC估测  

SOC estimation based on gas-liquid dynamics model and FFRLS-EKF algorithm

作　　者：栗欢欢[1] 孙化阳 陈彪[1] 王亚平[2] LI Huanhuan;SUN Huayang;CHEN Biao;WANG Yaping(Automotive Engineering Research Institute of Jiangsu University,Zhenjiang Jiangsu 212013,China;School of Material Science&Engineering,Jiangsu University,Zhenjiang Jiangsu 212013,China)

机构地区：[1]江苏大学汽车工程研究院,江苏镇江212013 [2]江苏大学材料科学与工程学院,江苏镇江212013

出　　处：《电源技术》2021年第11期1435-1438,1481,共5页Chinese Journal of Power Sources

基　　金：江苏省自然科学基金面上项目(BK20201426);江苏省科技支撑计划重点项目(BE2019010)。

摘　　要：针对当前亟需解决锂离子电池荷电状态(SOC)估计在动态与稳态工况下无法同时保持高精度的问题,利用气液动力学电池模型(GLD)与递推最小二乘法和扩展卡尔曼滤波算法提出联合SOC估算算法,采用含遗忘因子的递推最小二乘法(FFRLS)对模型的参数进行在线辨识,以消除原始算法存在的估算误差波动问题。利用恒流和动态应力测试(DST)工况进行了仿真验证。与单独采用FFRLS的算法以及原始算法进行对比,结果表明,所提算法具有更高的估算精度,最大误差为2.62%,具有估算精确度高和鲁棒性强的优点。This paper proposes an estimation algorithm based on the gas-liquid dynamic battery model,combining the recursive least square method and the extended Kalman filter algorithm to solve the problem that the current state of charge(SOC)estimation of lithiumion batteries(LIBs)is not accurate enough in both dynamic and steady-state conditions.First,a lithium ion battery model based on forgetting factor recursive least square method(FFRLS)and extended Kalman filter(EKF)is established,which using the gas-liquid dynamic battery model(GLD).The forgetting factor recursive least square method(FFRLS)is used to identify the online parameters of the model,and eliminate the fluctuation of estimation error in the original algorithm.Simulation tests are carried out under constant current and Dynamic Stress Test(DST)conditions,and compared with the original model and that using simple FFRLS.The results show that the maximum estimation error of the algorithm proposed by this paper is 2.62%,which has the advantages of high estimation accuracy and strong robustness.

关 键 词：气液动力学电池模型 递推最小二乘法 扩展卡尔曼滤波 SOC估算 

分 类 号：TM912[电气工程—电力电子与电力传动]