Analysis of Early-Stage Behavior and Multi-Parameter Early Warning Algorithm Research for Overcharge Thermal Runaway of Energy Storage LiFePO4 Battery Packs

• Published in: Journal of the Electrochemical Society Vol. 171; no. 9; pp. 90531 - 90542
• Main Authors: Wang, Canxiong, Du, Jianhua, Ye, Xianghu, Wei, Senrong, Zheng, Suzhen, He, Xingfeng, Wang, Jiabin, Xiong, Leji, Ou, Yingjie, Tu, Ran
• Format: Journal Article
• Language: English
• Published: IOP Publishing 02.09.2024
• Subjects: early warning algorithm; LiFePO4 battery; multi-parameter; overcharge; thermal runaway
• ISSN: 0013-4651; 1945-7111
• DOI: 10.1149/1945-7111/ad7bf0

Overcharging of lithium-ion batteries may lead to severe thermal runaway (TR) incidents, resulting in significant economic losses and safety hazards. Therefore, it is crucial to research early warning methods for TR behavior in overcharged lithium batteries. This study initially conducted overcharging experiments on LiFePO4 battery packs under different initial charging states and charging rates, analyzing variations in temperature, voltage, and inter-group pressure during overcharging. The TR process was divided into three stages: non-overcharged, early, and middle. Based on this, temperature change rate, pressure change rate, and voltage were extracted as input feature parameters, and the Mean Shift algorithm was employed for stage identification and classification of overcharging experiments on LiFePO4 battery packs. According to experimental results, the algorithm achieved an accuracy of over 96% in stage identification and classification of TR in overcharged lithium batteries, accurately determining the current stage of TR and providing a reliable and effective solution for preventing TR in overcharged lithium batteries. Simulated the thermal runaway(TR) of a single cell in the lithium-ion battery pack. Analyzed variations in temperature, voltage, and inter-group pressure during the overcharging process. Developed an early warning algorithm for TR in LiFePO4 battery packs based on multiple parameters. This algorithm can accurately judge the charging state of the battery.