A multi-mobile robot cooperative method is proposed to overcome the problem of low inspection efficiency in the large substation, by using the improved biological excitation neural network algorithm in this paper. Firstly, the shortcoming of the biological excitation neural network algorithm is analyzed, such as tortuous paths, large turning angle, and then an improved method is proposed. Combining the improved method and Hungarian algorithm, the cooperative inspection task of multi-robot for multi-task points is completed. Then, a variable tangent method is designed to decompose the substation area into several sub-regions without obstacles, and the priority heuristic algorithm is proposed for the robots to complete the full-area inspection task, meanwhile the improved method also be used for the robots to jump out the dead zone. Simulation experiments show that compared with the original algorithm and A* algorithm, the improved algorithm has obvious optimization effects in terms of path length and turning times, and the feasibility of the multi-robot collaborative inspection scheme is also proved throught the simulation.