The intelligent perception of overheating of insulated jumper clamp under high temperature and high load conditions in summer is taken as the research object, and the three-dimensional finite element model of electrical thermal multi-physical field coupling of insulated jumper clamp under typical working conditions is established. The validity of the model is verified by experiments, and the temperature field distribution data under different current load, sunlight intensity, environmental temperature and wind speed factors are obtained as jumper clamp overheating perception model sample. The reverse learning strategy is introduced to improve the ability of sparrow search algorithm (SSA)in global search and improved sparrow search algorithm(ISSA) is established. ISSA optimized BP neural network (ISSA-BPNN) is used to establish the temperature prediction model of insulated jumper clamp, and the prediction accuracies of ISSA-BPNN, particle swarm optimization BP neural network (PSO-BPNN), genetic algorithm optimized BP neural network (GA-BPNN), sparrow search algorithm BP neural network (SSA-BPNN) and BP neural network are evaluated using the mean square value and determination coefficient. The results show that compared with the prediction models of the other four algorithms, the ISSA-BPNN model can control the average prediction error within 0.71%, with higher prediction accuracy and faster convergence speed. It can more accurately predict insulated jumper clamp temperature rise, providing a basis for the state detection and evaluation of the insulated jumper clamp.