重庆理工大学学报2025，Vol.39Issue(7)：50-58,9.DOI:10.3969/j.issn.1674-8425(z).2025.04.007
融合极化特性和LSTM模型的电池温度估计
Battery temperature estimation through integrating polarization characteristics and LSTM model
摘要
Abstract
Data-driven approaches,especially long short-term memory(LSTM)neural network,the lack of theoretical guidance on model training strategies undermines the accuracy and development efficiency of the models.To address the issue,a series of battery performance test experiments are firstly conducted on a battery test platform.Then,the temperature characteristics of the battery are analyzed at different rates and ambient temperatures.Next,the polarization characteristics and heat production characteristics of the battery are analyzed at different rates and temperatures.It is found the temperature estimation model is divided into low,medium,and high-rate conditions.Thus,this paper proposes a battery temperature estimation strategy based on LSTM considering polarization characteristics under wide temperatures.Finally,the accuracy and generalization ability of the model are tested at different ambient temperatures and discharge rates.Results show the ME and MAE of the LSTM temperature estimation model considering polarization characteristics are 0.98 ℃ and 0.11 ℃ respectively.Meanwhile,a comparison with the traditional LSTM model suggests the ME and MAE are reduced by 1.85 ℃ and 0.76 ℃ respectively,and the training time is cut by 41.81％,all of which indicate the temperature estimation model based on LSTM considering polarization characteristics achieves higher accuracy and training efficiency.关键词
锂离子电池/温度估计/极化特性/LSTMKey words
lithium-ion battery/temperature estimation/polarization characteristics/LSTM分类
动力与电气工程引用本文复制引用
刘良,许光光,盘朝奉,王丽梅..融合极化特性和LSTM模型的电池温度估计[J].重庆理工大学学报,2025,39(7):50-58,9.基金项目
国家自然科学基金项目(52072155) （52072155）
