Multi-criteria Integrated Early Warning of Thermal Runaway Risk

With the advancement of new energy storage technologies and their widespread industrial applications, the issue of thermal runaway in lithium battery energy storage systems has become increasingly significant. Thermal runaway in energy storage systems can not only result in equipment damage and extended downtime but also pose serious threats to personnel safety and the environment. Therefore, early warning of thermal runaway in energy storage systems has gained paramount importance and has garnered extensive research attention among domestic scholars. However, conventional methods for thermal runaway prediction primarily rely on empirical models, lacking a comprehensive analysis and profound understanding of operational data from energy storage systems. Data-driven approaches offer a fresh perspective on thermal runaway prediction, enabling the identification of potential risk factors from extensive historical data and providing accurate warning and prevention strategies. This paper first introduces the principles and methods of data-driven algorithms, while exploring how abundant historical data can be utilized to identify latent thermal runaway risk factors. A comparison with traditional warning methods is also provided, demonstrating that the approach proposed in this paper outperforms in terms of false alarms and missed warnings. Finally, the paper discusses the potential challenges and future directions in proactively addressing thermal runaway risks through data-driven prevention.