Week-level early warning strategy for thermal runaway risk based on real-scenario operating data of electric vehicles

Abstract

Effective detecting thermal runaway risk in batteries are crucial for the rapid development and widespread adoption of electric vehicles. In this study, a strategy based on signal analysis is developed to realize the early warning of battery thermal runaway risk at the weekly level, without being limited by battery material systems. Firstly, a longitudinal outlier average method is developed to quantify the potential risk of thermal runaway in batteries and compared with a preset threshold to identify cells with performance anomalies. Secondly, an alarm assessment mechanism is developed, which integrates ongoing and historical operating data of suspicious cells across multiple decision layers. By employing an improved information entropy weighting method, this mechanism provides a comprehensive assessment of battery pack consistency, addressing issues related to false alarms and sporadic alerts. Finally, the effectiveness of this strategy is validated through actual vehicles involved in thermal runaway.