基于粒子滤波的锂电池SOC二次滤波算法

Estimation of lithium battery SOC based on KPDF algorithm

荷电状态(State-of-Charge,SOC)估计作为电池管理系统的核心算法,是电池状态估计与保护控制的重要依据。基于改进PNGV模型,应用带有遗忘因子的递推最小二乘算法实现电池模型参数实时更新;在此基础上,提出了将粒子滤波与卡尔曼滤波相结合的SOC二次滤波算法,结合系统状态方程对粒子滤波结果进行卡尔曼二次滤波,从而在计算量接近粒子滤波的前提下提高估计精度,克服了拓展卡尔曼滤波算法必须对实际工况进行线性化,以及粒子滤波算法在信噪比较低时粒子不能准确描述后验概率而导致结果偏移的问题。通过电池实验仿真验证了模型参数识别以及SOC估计算法的优越性。

As the core algorithm of battery management system,the estimation of SOC is an important basis for battery state estimation and system＇s protection and control. Based on the improved PNGV model,this paper uses the forgetting-factor recursive least square method to update the battery model parameters in real time. On this basis,the SOC dual filter algorithm is proposed which combined the PF and KF. It performs Kalman filtering after particle filtering to realize accuracy SOC estimation. The dual filter algorithm solves the problem of nonlinear error of EKF and the problem of result migration of PF in the case of low SNR. Finally,the superiority of battery model parameters and SOC estimation algorithm are verified through battery experiment and simulation.