基于GWO-BP神经网络的电池SOC预测方法研究

RESEARCH ON PREDICTION METHOD OF BATTERY SOC BASED ON GWO-BP NETWORK

为了进一步提高传统BP神经网络在电动汽车电池SOC预测中的精度,基于电动汽车云平台的以10 s为采样周期的电池运行数据,对电动汽车电池SOC的预测方法进行研究。对云平台数据进行预处理,选择电池包电压、电流和平均温度作为神经网络的输入;同时根据电流的特点,分别建立充电和放电过程的SOC神经网络预测模型;采用一种基于灰狼优化算法的BP神经网络(GWO-BP),通过灰狼优化算法优化BP神经网络的初始权值和阈值,从而减小神经网络预测误差;利用2503组云平台数据对未优化的BP神经网络、GWO-BP神经网络和粒子群优化的BP神经网络(PSO-BP)进行测试实验。实验结果表明,GWO-BP神经网络的放电过程平均误差为0.48%,充电过程平均误差为0.34%,明显高于未优化的BP神经网络和PSO-BP神经网络,具有较高的预测精度。

In order to further improve the accuracy of traditional BP neural network in the prediction of EV battery SOC,the prediction method of EV battery SOC is studied based on the battery operation data of the EV cloud platform with the sampling period of 10 seconds.The data of cloud platform was preprocessed,and the voltage,current and average temperature of battery pack were selected as the input of neural network.At the same time,according to the characteristics of current,the SOC neural network prediction model of charging and discharging process was established.A kind of BP neural network(GWO-BP)based on grey wolf optimization was used to optimize the initial weight and threshold value of BP neural network,so as to reduce the prediction error of neural network.2503 groups of cloud platform data were used to test BP neural network,GWO-BP neural network and PSO-BP neual network.The test results show that the average error of GWO-BP neural network is 0.48%in the discharge process and 0.34%in the charging process,which is significantly higher than the traditional BP neural network and PSO-BP neural network,and has higher prediction accuracy.