基于改进无迹卡尔曼滤波锂电池SOC估计

SOC Estimation of Lithium-ion Battery Based on Improved Untraced Kalman Filtering

锂电池的荷电状态(SOC)估计一直是电动汽车技术的重要研究方向。在实际应用中,准确估计电池SOC不仅可延长电池寿命,提高能源利用效率,还可避免过充和过放等安全问题。基于二阶RC等效电路模型,通过遗忘因子递推最小二乘法(FFRLS)对模型进行参数辨识,在UDDS工况下阐述了四种卡尔曼滤波衍生算法,经过实验对比得到最优SOC估计算法。实验结果表明,多新息无迹卡尔曼滤波算法将系统状态单新息转换为历史状态估计矩阵,SOC估计过程中平均误差控制在0.73%左右,在复杂系统工况下具有较高的估计精度和鲁棒性能。

The estimation of state of charge(SOC)of lithium-ion battery has always been an important research direction of electric vehicle technology.In practical applications,accurate estimation of battery SOC can not only extend battery life,improve energy efficiency,but also avoid safety problems such as overcharge and overdischarge.In this paper,based on the second-order RC equivalent circuit model,parameters of the model were identified by the forgotten factor recursive least square method(FFRLS),and four types of Kalman filtering derived algorithms were described under the urban dynamometer driving schedule(UDDS).The optimal SOC estimation algorithm was obtained through comparative experiment,whose results showed that the multi-new-information untraced Kalman filtering(MIUKF)algorithm converts system state information into historical state estimation matrix,and achieves an average error steadily controlled around 0.73%during SOC estimation process,demonstrating its high estimation accuracy and robustness in complex system working conditions.