基于WOA-BP神经网络估算锂离子电池SOC

Estimating SOC of Li-ion battery based on WOA-BP neural network

准确的荷电状态(SOC)估算可为电动汽车的可靠运行提供安全保障。提出将鲸鱼优化算法(WOA)和BP神经网络相结合的锂离子电池SOC估算方法。电池模型采用一阶RC电路,基于遗忘因子递推最小二乘法对模型参数进行辨识,通过电池实际状况自适应地调整校正,并采用WOA-BP神经网络算法,克服BP神经网络易陷入局部极小值和收敛速度慢的难点。与传统BP神经网络算法相比,基于WOA-BP的SOC估算方法,平均绝对误差降低1.9%,均方根误差减小4.1%,表明具有更高的鲁棒性和精确性。

The actual estimation of state of charge(SOC)could provide security for the reliable operation of electric vehicles.A Li-ion battery SOC estimation method based on a combination of whale optimization algorithm(WOA)and BP neural network was proposed.A first-order RC circuit was used in the battery model,the parameters of the model were identified based on the forgetting factor recursive least squares method to adjust the correction adaptively through the actual battery condition.The WOA-BP neural network algorithm was used to overcome the difficulties of BP neural networks which tended to fall into local minimum and slow convergence speed.Compared with the traditional BP neural network algorithm,the mean absolute error of the WOA-BP neural network algorithm was reduced by 1.9%and the root mean square error was reduced by 4.1%,indicating it had higher robustness and accuracy.