基于自适应FFRLS和改进CEKF锂电池SOC的估算

Estimation of lithium battery SOC based on adaptive FFRLS and improved CEKF

锂离子电池荷电状态(SOC)的准确估计是电池管理系统的重要核心技术之一,也是延长电池寿命的关键。但是SOC的准确实时估计困难,且精度不高。选取以三元锂电池为研究对象,针对EKF在线性化过程中产生的非线性误差,提出改进的补偿扩展卡尔曼算法(compensation for extended Kalman,CEKF)。该算法用GA优化的BP神经网络预测k时刻EKF的非线性误差等,从而补偿扩展卡尔曼k时刻的非线性误差等,且用自适应FFRLS对模型参数进行参数辨识,以DST和BBDST进行实验验证。实验结果表明,该算法估算SOC的精度范围在2%左右,且最大误差和平均误差都比EKF小得多,能更加有效追踪SOC的理论值,且该算法估计的SOC稳定性也比EKF稳定。

The accurate estimation of state of charge of lithium-ion batteries is one of the important core technologies of battery management system,and also the key to prolong battery life.However,the accurate real-time estimation of SOC is difficult and the accuracy is not high.Taking ternary lithium battery as the research object,an improved compensation for extended Kalman algorithm was proposed to solve the nonlinear error of EKF in the process of linearization.In this algorithm,GA-optimized BP neural network was used to predict the nonlinear error of EKF at time k,so as to compensate the nonlinear error at time k of the extended Kalman.Moreover,the model parameters were identified by adaptive FFRLS.DST and BBDST were used for experimental verification.The experimental results show that the accuracy range of the proposed algorithm is about 2%,and the maximum error and average error are much smaller than that of EKF,so it can track the theoretical value of SOC more effectively,and the estimated SOC stability of the proposed algorithm is also more stable than that of EKF.