Dueling DQN-based hyper-heuristic algorithm for solving the location routing problem

In order to solve the vehicle path siting problem with capacity in the logistics system more efficiently and make the system realize the overall optimization, so a hyper-heuristic algorithm based on Dueling DQN is proposed. The algorithm uses the Dueling DQN algorithm as the high-level selection strategy of the hyper-heuristic algorithm, evaluates the advantages and disadvantages of the underlying operators, and utilizes the reward and punishment values to weigh the operators; meanwhile, the simulated annealing mechanism with rewarming operation is used as the acceptance criterion of the algorithm, so as to control and guide the algorithm to search for the solution space more effectively. In addition, encodings are constructed that guarantee the feasibility of the solution and can avoid the use of solution repair techniques. Finally, a standardized algorithm for the capacitated siting path problem is computed and compared with other algorithms to verify the effectiveness and robustness of the algorithm on this problem.