摘要
锂离子电池对于推动可再生能源的广泛利用、实现电网的平稳运行等具有举足轻重的意义,但其热失控形成的热蔓延会造成巨大损失。锂离子电池热失控预警技术已经成为预防电池安全问题的有效手段。以280Ah方形铝壳磷酸铁锂(LFP)电池为例,本工作搭建了锂离子电池热失控试验平台,采用正常循环与过充诱发热失控的方式,采集了锂电池表面温度、输出电压及电池膨胀压力信号等数据集,样本总量为66895个,其中过充热失控样本245个。该数据集在时间序列上描述了电池从正常循环到过充再到热失控的连续过程,通过测量电池膨胀压力信号,有效映射了电池内部体积变化。进一步在数据集的基础上,设计了一种基于注意力机制-残差结构-长短时记忆网络(SE-Res-LSTM)的电池膨胀压力回归预测算法,并利用预测误差对电池过充热失控进行实时检测。在时效性上,在电池发生过充12 s后,即能检出早期过充热失控风险,相比于以电池表面达到60℃的温度划分方法提早233s,显著提高了预警时效性和准确性。
Lithium-ion batteries play a crucial role in promoting the widespread use of renewable energy and ensuring the stable operation of power grids.However,thermal runaway-induced thermal spread in these batteries can result in significant losses.Therefore,thermal runaway warning technology for lithium-ion batteries is essential for preventing safety issues.This study develops a lithium-ion battery thermal runaway test platform,collecting data on surface temperature,output voltage,and battery expansion pressure during normal cycles and overcharge-induced thermal runaway events.A total of 66895 samples were obtained,including 245 samples exhibiting overcharge thermal runaway.The dataset captures the continuous transition from normal operation to overcharge and subsequent thermal runaway in a time series.The internal volume changes of the battery are effectively represented by measuring the expansion pressure signals.A regression prediction algorithm based on SE-Res-LSTM is developed to predict battery expansion pressure,utilizing the prediction error to detect overcharge-induced thermal runaway in real-time.In terms of timeliness,early detection of overcharge thermal runaway is achieved 12 s after the onset of overcharging,which is 233 s earlier than the conventional temperature-based method that triggers a warning when the battery surface reaches 60℃.This significantly enhances the accuracy and responsiveness of early warning systems.
作者
蒋建杰
楼平
徐国华
来骏
王瑶
曹志成
张炜鑫
曹元成
JIANG Jianjie;LOU Ping;XU Guohua;LAI Jun;WANG Yao;CAO Zhicheng;ZHANG Weixin;CAO Yuancheng(Huzhou Electric Power Design Institute Co.,Ltd,Huzhou 313098,Zhejiang,China;State Grid Huzhou Electric Power Supply Company,Huzhou 313099,Zhejiang,China;School of Electrical and Electronic Engineering,Huazhong University of Science and Technology,Wuhan 430074,Hubei,China)
出处
《储能科学与技术》
CAS
CSCD
北大核心
2024年第11期4187-4197,共11页
Energy Storage Science and Technology
基金
浙江泰仑集团有限责任公司科技项目(HZJTKJ2022-04)。
关键词
锂离子电池
热失控
数据集
预测误差
早期预警
lithium-ion battery
thermal runaway
dataset
prediction error
early warning