基于LSTM-SVR的锂电池健康状态预测研究

Research on lithium battery state of health prediction based on LSTM-SVR

锂电池健康状态(SOH)是表征电池衰退的重要指标,也是新能源汽车运行寿命的一个重要指标。为了提高锂电池健康状态SOH的预测精度,采用深度学习LSTM与SVR结合的算法来对锂电池健康状态进行预测,并利用网格搜索(GS)搜索SVR超参数。首先,使用平均放电电压、平均放电温度、容量作为健康因子(HI);其次,利用以往锂电池数据集对算法进行验证。实验结果表明:使用LSTM-SVR算法相比于LSTM算法在RMSE指标和拟合程度上更优,其均方根误差在0.6以内,平均绝对百分误差在0.6%以内。

Lithium battery's state of health(SOH)is an important indicator of battery decline and an important indicator of the operating life of new energy vehicles.In order to improve the prediction accuracy of lithium battery health status SOH,this paper proposed the algorithm of deep learning LSTM and SVR to predict the SOH of lithium battery,and use grid search(GS)to search SVR hyper-parameters.First,the average discharge voltage,average discharge temperature,and capacity were used as the health factor(HI).Second,the algorithm was verified using the lithium battery data set of the NASA Research Center.The experimental results show that the LSTM-SVR algorithm is better than the LSTM algorithm in the RMSE index and the degree of fitting,its root mean square error is within 0.6,and the average absolute percentage error is within 0.6%.