基于GNL模型自适应无迹卡尔曼滤波的电动汽车的荷电状态估计

随着电动汽车的高效发展，逐年递增的退役动力电池回收利用已刻不容缓，对电池进行精确、可靠的荷电状态(state of charge，SOC)估计是实现电池梯次利用的关键技术。传统估计方法均未考虑对老化电池影响较高的自放电因素，本文采用在二阶RC模型基础上考虑了自放电因素的GNL电路等效模型，通过脉冲放电对模型参数进行辨识。对相应的状态空间方程利用矩阵二次型方法进行离散化，并利用自适应无迹卡尔曼滤波算法对SOC进行实时估计及更新。在间歇恒流工况和变电流工况下以老化电池为实验对象对算法进行了对比验证，结果表明双卡尔曼滤波法在初值估计不准确的时候不能及时收敛到SOC真值附近并跟随，基于二阶RC模型的自适应滤波算法估计的误差在工况后期较大，基于GNL模型的自适应滤波算法对老化电池的估计精度较高，误差在0.5%之间。结果表明该方法可使状态估计值具有较小的误差和快速跟随性，满足了SOC 估计的实际需求。

Electric Vehicle Battery SOC Estimation Based on GNL model Adaptive Kalman Filter

With the high efficiency development of electric vehicle, it is urgent to recycle and utilize the decommissioned power battery .Accurate and reliable state of charge state estimation is the key technology to realize ladder utilization of battery. The traditional estimation methods do not take into account the high self-discharge factor that affect the aging battery. This paper adopts the GNL circuit equivalent model considering the self-discharge factor based on the second-order RC model .parameter identification is finished through pulse discharge experiment. The state space equation is discretized by matrix quadratic method, and the adaptive unscented Kalman filter algorithm is used to estimate and update the SOC (state of charge). Under the condition of intermittent constant current and variable current, the algorithms were compared and verified with aging battery as experimental object. The results show that the double Kalman filter cannot converge to the true value of SOC (state of charge) in time when the initial value is not accurate. The estimation error of adaptive filtering algorithm based on second-order RC model is large in the late stage of working condition and the max error of adaptive filtering algorithm based on GNL model is 0.5%. The results show that the proposed method can make the state estimation have less error and fast follow, and meet the actual demand of SOC (state of charge) estimation.