A Method of the Traveling Salesman Problem Based On Q-Learning Reinforcement Learning and Local Dynamic Programming

This study explores optimizing the Traveling Salesman Problem (TSP) using Q-Learning reinforcement learning. The proposed method builds a Q-table to learn the optimal path and employs dynamic programming for local optimization. Experimental results demonstrate significant optimization effects of this method across multiple test cases, showing higher efficiency and precision compared to traditional heuristic algorithms and genetic algorithms, thus exhibiting promising application prospects. This research offers a novel perspective and method for addressing real-world TSP problems.