电池2025，Vol.55Issue(6)：1270-1276,7.DOI:10.19535/j.1001-1579.2025.06.008
基于时间卷积网络的锂离子电池SOC估计
SOC estimation of Li-ion batteries based on temporal convolutional network
摘要
Abstract
Due to the insufficient accuracy of current state of charge(SOC)estimation methods,a method based on temporal convolutional networks(TCN)is proposed.A TCN model is constructed,using voltage,current and temperature data during the battery operation process as input features.Modeling and validation are carried out under conditions of 5 ℃,10 ℃,15℃,20 ℃and 25 ℃.To evaluate the model's performance,it is compared with recurrent neural networks(RNN),long short-term memory networks(LSTM)and backpropagation neural networks(BPNN),using metrics such as mean squared error(MSE),root mean squared error(RMSE)and mean absolute error(MAE).The TCN shows high SOC estimation accuracy under different temperatures,especially in the room temperature range(20 ℃ and 25 ℃),with the lowest MAE reaching 0.16％,significantly outperforming other comparison models.It also exhibits strong prediction stability and good resistance to fluctuations.The TCN demonstrates excellent accuracy and robustness under multiple operating conditions,with a good application effect.关键词
锂离子电池/荷电状态(SOC)估计/时间卷积网络(TCN)/温度Key words
Li-ion battery/state of charge(SOC)estimation/temporal convolutional network(TCN)/temperature分类
信息技术与安全科学引用本文复制引用
夷文玉,眭祥,刘进福..基于时间卷积网络的锂离子电池SOC估计[J].电池,2025,55(6):1270-1276,7.基金项目
江苏省自然科学基金(BK20220241),江苏省高等学校自然科学面上基金(23KJD460001),江苏高校青蓝工程优秀青年骨干教师项目 （BK20220241）
