异常情况下电动汽车锂电池SOC估计

Li-Battery SOC Estimation of Electric Vehicle Under Abnormal Conditions

为提高锂电池在状态突变、模型不准确、SOC初始误差大等异常情况下的SOC估计精度和收敛速度,提出了基于强跟踪卡尔曼滤波算法的SOC估计方法。建立了锂电池的双RC等效电路模型,使用HPPC方法辨识了模型参数;分析了扩展卡尔曼滤波原理和缺陷,在误差协方差矩阵中引入时变渐消因子,用于改进修正系数矩阵,强行使残差序列保持正交特性,基于此原理提出了强跟踪卡尔曼滤波算法。经仿真验证,在模型不准确和状态突变情况下,强跟踪卡尔曼滤波的最大估计误差为2%,而扩展卡尔曼滤波最大误差为4.5%;在SOC初始误差较大情况下,强跟踪卡尔曼滤波在15 s内收敛至真值,而扩展卡尔曼滤波在40 s时收敛至真值。

In order to improve SOC estimation accuracy and rate of convergence of lithium battery in abnormal conditions of inaccurate model, mutation status, and large initial error, etc., a SOC estimation method based on strong tracking Kalman filter is proposed. Double RC equivalent circuit model of lithium battery is built, and model parameters are identified by HPPC approach. Principle and defects of extended Kalman filter are analyzed. Time-varying fading factor is introduced to error covariance matrix, to modify correction coefficient matrix, which make residual sequence orthogonal. Based on this principle, strong tracking Kalman filter algorithm is proposed. Simulation verifies that, in condition of inaccurate model and mutation status, the maximum estimation error by strong tracking Kalman filter is 2%, and 4.5% by the extended Kalman filter. In condition of large SOC initial error, it takes 15 s for the strong tracking Kalman filter, and 40 s for the extended Kalman filter to converge to the true value.