基于修正协方差近似二阶扩展卡尔曼滤波算法的电池荷电状态估算  

Estimation of Battery Charge State Based on MVASOEKF Algorithm

作　　者：王伯运 何耀[1] 郑昕昕[1] Wang Boyun;He Yao;Zheng Xinxin(Hefei University of Technology,Hefei 230009)

机构地区：[1]合肥工业大学,合肥230009

出　　处：《汽车工程师》2023年第2期1-8,共8页Automotive Engineer

基　　金：中国南方电网集团广东电科院能源技术有限责任公司委托项目(03872020000374CG)。

摘　　要：为解决扩展卡尔曼滤波(EKF)算法中由非线性变换忽略高阶泰勒项引起的荷电状态(SOC)估算误差和在迭代过程中协方差容易出现病态的问题,采用修正协方差近似二阶扩展卡尔曼滤波(MVASOEKF)算法,通过混合脉冲功率特性试验对等效模型内部参数进行离线辨识并建立了二阶RC等效电池模型,在MATLAB/Simulink平台上进行SOC估算,结果表明,EKF算法估算平均绝对误差约为2.0%,MVASOEKF算法估算平均绝对误差约为0.5%,与EKF算法相比,MVASOEKF算法虽增加了一定的计算量,但是SOC估计精度明显改善,且收敛性更好。In order to solve the problem of State Of Charge(SOC)estimation error and ill-conditioned covariance in the iteration process caused by nonlinear transformation ignoring the high-order Taylor term in Extended Kalman Filter(EKF)algorithm,a Modified Covariance Approximate Second-Order Extended Kalman Filter(MVASOEKF)algorithm was adopted.Through the hybrid pulse power characteristic experiment,the internal parameters of the equivalent model were identified offline and the second-order RC equivalent battery model was established.The SOC was estimated on MATLAB/Simulink platform.The results show that the average absolute error value of EKF algorithm is about 2.0%,and the average absolute error value of MVASOEKF algorithm is about 0.5%.Compared with EKF algorithm,although MVASOEKF algorithm has more computation amount,but the SOC estimation accuracy has been significantly improved,and the convergence is better.

关 键 词：二阶RC 等效电池模型 参数辨识 荷电状态估算 修正协方差近似二阶扩展卡尔曼滤波 

分 类 号：TM911[电气工程—电力电子与电力传动]