A new method for thermal runaway warning of electric vehicle charging

Abstract With the large-scale development of electric vehicle (EV) in China, the frequent occurrence of EV fire accidents has attracted the attention of insiders to the fault monitoring and early warning during the charging process of EV. According to data from the National Emergency Management Ministry, there were over 3,000 EV fire accidents nationwide in 2023, underscoring the urgency for enhanced fault monitoring and early warning measures. In response to these issues, this article proposes an EV thermal runaway early warning method based on Temporal Convolutional Network (TCN) and Gated Recurrent Unit (GRU) by combining the two technologies. The method aims to improve EV safety by monitoring battery status for early thermal runaway warning. First, the historical normal charging data of the battery are preprocessed and feature extracted using the TCN, and then features fed into GRU for time-series modeling and forecasting. This combined model not only demonstrates high prediction accuracy and stability but also swiftly responds to abnormal conditions during charging, effectively preventing thermal runaway accidents and ensuring charging safety. Furthermore, this model possesses excellent generalization capabilities and can adapt to different types and specifications of EV battery systems.