基于FF-MILS和UKF算法的锂电池SOC估算  

Estimation of state of charge for lithium-ion battery based onmultiinnovation recursive least square algorithmand unscentedKalman filter

作　　者：黄敬尧[1] 李凌峰 张扬 宋轩宇 HUANG Jingyao;LI Lingfeng;ZHANG Yang;SONG Xuanyu(School of Electricity and New Energy,China Three Gorges University,Yichang Hubei 443002,China;Hubei Energy Group Ezhou Power Generation Co.,Ltd.,Ezhou Hubei 436000,China)

机构地区：[1]三峡大学电气与新能源学院,湖北宜昌443002 [2]湖北能源集团鄂州发电有限公司,湖北鄂州436000

出　　处：《电源技术》2021年第6期711-715,735,共6页Chinese Journal of Power Sources

摘　　要：为了建立精确的锂离子电池模型,在线监测电池的荷电状态(SOC),采用二阶等效电路模型,通过实验数据拟合开路电压(OCV)与荷电状态的对应关系。选用遗忘因子多新息递推最小二乘法(FF-MILS)为在线辨识算法完成对锂离子电池在线模型参数估测,同时将所得模型参数传入卡尔曼滤波器,完成对动力电池荷电状态的估算。在MATLAB/Simulink中实现该算法的编程,把电池综合测试仪对锂离子电池采样所得电流、电压、容量等实验数据导入算法进行仿真。结果表明,该算法迅速收敛初值误差,并在稳定状态下最大误差不超过2%,从而验证了该算法的有效性以及对外界干扰的鲁棒性,可以用来实现对车用锂离子电池状态的准确估算。An accurate model of lithium-ion battery was developed for real-time and effective monitoring of battery status.A second order equivalent circuit model was used to fit the corresponding relationship between open circuit voltage(OCV)and charge state through experimental data the forgetting factor multi-innovation least squares(FFMILS)algorithm was introduced to establish the parameters of battery model.Then,the state of charge(SOC)of the battery could be estimated by using Kalman filter with model parameters.A comprehensive tester was used to sample the current,voltage,capacity and other data of the battery for simulation.The results indicate that the initial error of the algorithm is rapidly converged,and the maximum error is less than 2%during stable state,which verifies the superiority of the algorithm.Whereupon it is promising in applications of accurate estimation for the status of vehicle lithium-ion battery.

关 键 词：锂离子电池 荷电状态 递推最小二乘法 无迹卡尔曼滤波 在线参数辨识 

分 类 号：TM912[电气工程—电力电子与电力传动]