Optical Cable Path Planning Algorithm Based on Deep Reinforcement Learning

Traditional automatic path planning algorithm has the problem of missing data of optical cable vector length, which leads to the slow search speed of the algorithm. Combined with the actual withstand voltage level of power grid, the best path structure information features are extracted, the complex object input module is established, the optical cable vector length database is generated by using the random forest principle, and the three-dimensional import model of optical cable line is constructed to describe the tightness between nodes and adjacent nodes, query the actual physical path of nodes, and set the path automatic planning mode based on deep reinforcement learning. Experimental results: the average search speed of the two existing automatic path planning algorithms and the designed automatic path planning algorithm is 23.224m/s, 19.978m/s, 12.105m/s respectively, which proves that the automatic path planning algorithm combined with deep reinforcement learning has good application effect.