CCTEG Chinese Institute of Coal Science
CCTEG Taiyuan Research Institute
Shanxi Tiandi Coal Mining Machinery Co., Ltd.

电池健康状态(SOH)是锂离子电池的一项重要指标。为提高预测精度,提出了一种基于深度卷积神经网络(CNN)和长短期记忆网络(LSTM)的并行CNNLSTM网络模型,用于预测矿用锂电池的健康状况。该方法利用CNN获取数据局部特征,LSTM获取时间序列信息。然后将CNN层和LSTM层获取的信息合并为一个张量,输入额外的LSTM层,进一步获取信息,完成电池健康状态预测。通过对电池的放电容量、放电时间、内阻等特征进行选择和分析,验证了该模型能够有效地预测电池的健康状况。仿真结果表明,该模型在数据集上的预测误差均小于3%,均方根误差(RMSE)和平均绝对误差(MAE)值的平均值在0.484%和0.278%以内。

State of battery health (SOH) is an important indicator of lithiumion batteries. To improve predictionaccuracy, a parallel CNNLSTM network model based on deep convolutional neural network (CNN) and long shortterm memory network (LSTM) was proposed for predicting the health state of mining lithium batteries. This methodutilizes CNN to obtain local features of the data, while LSTM obtains time series information. Then, the informationobtained from the CNN layer and LSTM layer was merged into a tensor, and an additional LSTM layer was input tofurther obtain information and complete the prediction of the state of battery health. By selecting and analyzing thedischarge capacity, discharge time, internal resistance and other characteristics of the battery, it was verified thatthe model can effectively predict the health status of the battery. The simulation results showed that the predictionerror of the model on the dataset was less than 3%, and the average values of root mean square error (RMSE) andmean absolute error (MAE) were within 0. 484% and 0. 278% .

常映辉,王大钟,冀鹏飞,等.基于并行CNN-LSTM的矿用磷酸铁锂电池SOH预测[J].煤矿机电,2023,44(4):6-11.doi:10.16545/j.cnki.cmet.2023.04.002