基于滚动时域估计EKF算法的电池SOC预测  

Battery SOC estimation based on moving horizon estimation EKF algorithm

作　　者：李凌锋 陶晋宜[1] 张富强 LI Ling-feng;TAO Jin-yi;ZHANG Fu-qiang(Institute of Electrical and Power Engineering,Taiyuan University of Technology,Taiyuan Shanxi 030024,China;School of Electronic Engineering,Xidian University,Xi'an Shaanxi 710000,China)

机构地区：[1]太原理工大学电气与动力工程学院,山西太原030024 [2]西安电子科技大学电子工程学院,陕西西安710000

出　　处：《电源技术》2020年第7期991-994,共4页Chinese Journal of Power Sources

基　　金：工信部通信软科学研究课题(2018-R-21)。

摘　　要：扩展卡尔曼滤波算法(EKF)是分析电池SOC估计的有效工具,对于非线性系统具有良好的估计结果。但在实际情况下,测量噪声的模型不再是0均值白噪声,导致预测结果有较大误差。提出利用滚动时域(MHE)算法与扩展卡尔曼滤波算法结合,采用滚动时域窗口的思想,综合前N个时刻的窗口信息来估计当前的SOC。通过建立到达函数并加入M估计器替代测量误差,将预测问题转化为优化问题,并根据EKF的求解思想给出了近似最优解。实验仿真证明,该算法可以有效提高SOC的估计精度。Extended Kalman Filter(EKF)is an effective tool for analyzing battery SOC estimation and has good results for nonlinear systems.However,under actual conditions,the model for measuring noise is no longer a zero-mean white noise,resulting in a large error in the prediction results.In this paper,the Moving Horizon Estimation(MHE)algorithm is combined with the extended Kalman filter algorithm,and the idea of rolling time domain window is used to synthesize the window information of the first N times to estimate the current SOC.By establishing the arrival function and adding the M-estimator to replace the measurement error,the prediction problem is transformed into the optimization problem,and the approximate optimal solution is given according to the idea of EKF.Experimental simulations show that the algorithm can effectively improve the estimation accuracy of SOC.

关 键 词：SOC估算 EKF算法 MHE算法 M估计器 优化问题 

分 类 号：TM912[电气工程—电力电子与电力传动]