基于模糊无迹卡尔曼滤波算法的锂电池SOC估计  

SOC estimation of lithium battery based on fuzzy unscented Kalman filter

作　　者：安昌祖 张蕊萍[1] 张小周 刘浩[2] 董海鹰[1] AN Chang-zu;ZHANG Rui-ping;ZHANG Xiao-zhou;LIU Hao;DONG Hai-ying(Stool of Auomation and Elcrical Encring Lanzhou Jiong Uiversits,Lanzhou Cansu 70070 China;Tianshui Eetric Drive Reserth Istiute Co.Ltd,Tianshui Gansu 71000,China)

机构地区：[1]兰州交通大学自动化与电气工程学院,甘肃兰州730070 [2]天水电气传动研究所有限责任公司,甘肃天水741000

出　　处：《电源技术》2020年第3期333-336,356,共5页Chinese Journal of Power Sources

基　　金：大型电气传动系统与装备技术国家重点实验室开放基金(SKLLDJ042017005)。

摘　　要：针对未知噪声条件下在线估计锂电池荷电状态精度低的问题,提出了将无迹卡尔曼滤波算法与模糊推理相结合的模糊无迹卡尔曼滤波算法。为了验证算法的有效性,首先建立了适应于FUKF估计SOC的二阶电池模型,在此基础上,采用离线的参数辨识方法辨识模型中相应的参数并进行模型精确度验证,其次设计实验对比模糊无迹卡尔曼滤波方法与传统EKF、UKF方法的估算精度,实现FUKF方法精确度验证。实验结果表明在未知噪声条件下估算SOC,FUKF方法误差小于0.5%,EKF、UKF方法误差在0.5%~1%之间波动,FUKF方法较UKF方法具有收敛速度快、估算精度高的优点。Aiming at the problem of low accuracy of online estimation of State of Charge(SOC) of lithium batteries with unknown noise,a fuzzy unscented Kalman filter(UKF) combining with fuzzy reasoning is proposed.In order to verify the validity of the algorithm,first of all,it sets up suitable FUKF second order battery model to estimate the SOC,on this basis,by using the parameter identification method of offline recognition model and the corresponding parameters in model accuracy validation.Secondly,it designs experiments to compare fuzzy no trace Kalman filtering method with the traditional EKF,UKF methods estimate accuracy,and achieves precision FUKF method validation.The experimental results show that the FUKF method’s error is less than 0.5%,and the EKF and UKF methods’ error fluctuates between 0.5% and 1%,while the FUKF method has the advantages of faster convergence and higher estimation accuracy than the UKF method.

关 键 词：锂电池 SOC估计 模糊推理 模糊无迹卡尔曼滤波 参数辨识 

分 类 号：TM912[电气工程—电力电子与电力传动]