基于改进RLS算法的锂电池模型参数识别与SOC估计

Parameter Identification and SOC Estimation ofLithium Battery Model Based on Improved RLS Algorithm

SOC是反应锂电池剩余荷电状态的一个值.对电池荷电状态(SOC)的准确估算在电池管理系统(BMS)中意义重大.目前锂电池模型参数识别最常用的方法是递推最小二乘法(RLS,Recursive Least Square),但该算法中的固定遗忘因子很难满足锂电池的各种工况,导致参数识别在不同工况下的差异较大,不具有很好的适应性.因此,本文在二阶RC电路等效模型的基础上,采用一种可变遗忘因子参数识别VFFRLS和AEKF联合算法对锂电池SOC进行在线估计,从而减小由于参数辨识而引起的SOC估计误差.基于UDDS工况采用该联合算法对锂电池SOC进行估计,最后与RLS-AEKF算法以及单个AEKF算法进行比较,结果表明:VFFRLS-AEKF联合算法具有更高的准确性和稳定性.

SOC is a value reflecting the remaining charge state of lithium battery.Accurate estimation of battery state of charge(SOC)is of great significance in battery management system(BMS).At present,the most commonly used method for parameter identification of lithium battery model is Recursive Least Square(RLS),but the fixed forgetting factor in the algorithm is difficult to meet various working conditions of lithium battery,resulting in a large difference in parameter identification under different working conditions,which does not have good adaptability.Therefore,based on the second-order RC circuit equivalent model,a joint algorithm of variable forgetting factor parameter recognition VFFRLS and AEKF is used to estimate the SOC of lithium battery online,so as to reduce the SOC estimation error caused by parameter identification.In this paper,the JOINT algorithm is used to estimate the SOC of lithium battery based on UDDS conditions.Finally,compared with RLS-AEKF algorithm and single AEKF algorithm,the results show that the VFFRLS-AEKF joint algorithm has higher accuracy and stability.