Abstract
Based on ANN (artificial neural network), this paper puts forward an inspection path planning algorithm for substation robots, which combines BPNN (BP neural network) with nonlinear fitting and prediction ability with reinforcement Q learning algorithm with strong online learning ability. Firstly, the running environment of the robot is described and the environment type is determined. Secondly, given the state space description and action space description of the mobile robot, the input and output parameters of BPNN are determined. Then, the rules of obstacle avoidance for mobile robots are established, so that they can quickly jump out of the shock trap. The simulation results show that the convergence speed of the proposed algorithm is faster than that of APF (Artificial potential field) method, and the numerical value after convergence is stable and the fluctuation of the optimal solution is small. Moreover, both the path distance and the number of turns are superior to APF algorithm. It shows that the method proposed in this paper can help the inspection robot to traverse the inspection points and avoid collision targets.