In response to the problems of delayed early warning and high false alarm rate in the identification of abnormal temperature rise of the charging chassis of electric vehicles, this paper proposes a rapid identification method that integrates multiple features of temperature, temperature rise rate and spatial gradient. This method constructs a threedimensional feature system and establishes a hierarchical and multi-condition coupled identification rule to achieve early and accurate identification of local hotspots and sudden temperature rises. The actual parking space experiment verification shows that the proposed method can shorten the average warning time to 22 seconds, increase the identification accuracy to 96%, and significantly reduce the false alarm rate to less than 3% by relying on the spatial gradient feature, which has clear engineering application value.