基于自适应回归扩展卡尔曼滤波的电动汽车动力电池全生命周期的荷电状态估算方法

State of Charge Estimation Method of Electric Vehicle Power Battery Life Cycle Based on Auto Regression Extended Kalman Filter

该文针对传统扩展卡尔曼滤波(EKF)算法对电池数学模型精确的高度依赖与动态电池模型难以精确获得之间的矛盾问题,提出一种完全数据驱动的基于改进EKF算法的动力电池全生命周期荷电状态(SOC)估计方法。该方法为数据驱动的SOC估计方法和基于模型的SOC估计方法的良好结合,其优点在于:一方面抑制数据驱动方法存在累积误差的问题,并保留其良好的动态特性;另一方面改善基于模型的算法过度依赖电池模型的缺点,并保留其很好的鲁棒特性。该方法的创新之处在于将等效电路中难以获知的一部分视为以电池电流为输入,以内部电压为输出,以电池内部阻抗为时变参数的黑箱系统,并加以动态在线辨识,获得实时的动力电池真实状态,从而保证电池模型的准确性和动态性,真正实现动力电池全生命周期的SOC估算。仿真结果表明,该方法具有较好的鲁棒性和实用性。

The traditional extended Kalman filter(EKF)algorithm relies heavily on battery mathematic model and the dynamic battery model is difficult to obtain accurately.This paper proposes a completely data driven state of charge(SOC)estimation method based on improved EKF algorithm for battery life cycle.This method is a good combination of data-based SOC estimation method and model-based SOC estimation method.On the one hand,the SOC calculation method proposed in this paper can avoid the problem that data-driven SOC estimation method cannot eliminate the accumulated error and retain the data-driven algorithm good dynamic characteristics.On the other hand,it can improve the shortcoming of the model-based SOC estimation method that relies heavily on battery model while preserving the robustness.The hard-to-know part of the equivalent circuit is regarded as a black-box system,where the battery current is taken as input,the internal voltage as output and the internal impedance of the battery as time-varying parameter.The dynamic online identification is carried out to obtain the real state of the power battery,thereby ensuring the accuracy and dynamics of the battery model and realizing the battery life cycle SOC estimation.Simulation results show that the proposed method has good robustness and practicability.