In the process of power system operation, the overheating fault of power switch cabinet is the most common technical problem. This paper first analyzes the causes of the heating fault of the power switch cabinet, summarizes the physical relationship between the internal fault of the switch cabinet and the temperature of the cabinet, and establishes the relevant model of the internal heat flow of the switch cabinet. Through the temperature change feedback by the internal heat flow of the switch cabinet on the surface of the cabinet, the internal fault type of the switch cabinet can be effectively proved by the comparative analysis of the temperature cloud image. Therefore, the neural network algorithm is used to screen and analyze the thermal imaging cloud image of the switch cabinet surface, so as to judge the fault by comparison. This paper proves that the neural network algorithm combined with big data to compare and analyze the change of the surface cloud image of the switch cabinet can judge the fault type of the internal overheating fault of the switch cabinet, so as to realize the real-time monitoring of the power switch cabinet and improve the stability of the power system.