基于FFRLS和ASR-UKF滤波算法的锂电池SOC估计  

Lithium battery SOC estimation based on FFRLS and ASR-UKF filtering algorithm

作　　者：邓丹 刘胜永[1] 王顺利[2] 刘鹏辉 胡聪[3] DENG Dan;LIU Shengyong;WANG Shunli;LIU Penghui;HU Cong(School of Automation,Guangxi University of Science and Technology,Liuzhou Guangxi 545006,China;School of Information Engineering,Southwest University of Science and Technology,Mianyang Sichuan 621010,China;Guangxi Key Laboratory of Automatic Testing Technology and Instrument,Guilin University of Electronic Science and Technology,Guilin Guangxi 541004,China)

机构地区：[1]广西科技大学自动化学院,广西柳州545006 [2]西南科技大学信息工程学院,四川绵阳621010 [3]桂林电子科技大学广西自动检测技术与仪器重点实验室,广西桂林541004

出　　处：《电源技术》2024年第2期299-305,共7页Chinese Journal of Power Sources

基　　金：国家自然科学基金项目(No.62263001);广西自动检测技术与仪器重点实验室基金(No.YQ22203)。

摘　　要：锂电池在工作过程中,其内部参数易受多种因素影响,为提高锂电池在复杂环境下荷电状态(SOC)估计精度,以二阶戴维宁(Thevenin)等效模型为基础,结合遗忘因子递推最小二乘法(FFRLS)对模型参数进行在线辨识。针对传统卡尔曼滤波算法高度非线性及系统噪声不确定性等缺点,提出了一种自适应平方根无迹卡尔曼滤波(ASR-UKF)算法,该算法利用平方根算法处理均值和协方差,确保了状态协方差的半正定性和稳定性,并引入自适应滤波算法对噪声进行实时修正,消除了系统时变噪声影响。结果表明,FFRLS能有效解决数据饱和及算法矩阵计算量大的问题,等效模型精度高达98%。在混合动力脉冲特性(HPPC)测试和北京公交动态测试工况(BBDST)下,ASR-UKF算法SOC估计最大误差分别为3.264%和0.572%,具备更好的跟踪效果,验证了改进算法良好的收敛性与自适应性。In the working process of a lithium battery,its internal parameters are easily affected by many factors.To improve the accuracy of SOC estimation of lithium batteries in complex environments,the model parameters were identified online based on the second-order Thevenin equivalent model and the forgetting factor recursive least square method(FFRLS).Aiming at the high nonlinearity of the traditional Kalman filtering algorithm and the uncertainty of system noise,an adaptive square root unscented Kalman filtering(ASR-UKF)algorithm was proposed.The square root algorithm was used to process mean and covariance to ensure the semi-positive nature and stability of state covariance.An adaptive filtering algorithm was introduced to correct noise in real time to eliminate the impact of time-varying noise in the system.The results show that FFRLS can effectively solve the problem of data saturation and large calculation of the algorithm matrix,and the accuracy of the equivalent model is up to 98%.Under HPPC and BBDST working conditions,the maximum errors of SOC estimation of the ASR-UKF algorithm are 3.264%and 0.572%,respectively,which has a better tracking effect,verifying the good convergence and adaptability of the improved algorithm.

关 键 词：荷电状态 二阶Thevenin模型 遗忘因子递推最小二乘法 自适应平方根无迹卡尔曼滤波算法 

分 类 号：TM912.9[电气工程—电力电子与电力传动]