A New Method for Multi-Depot Vehicle Routing Problem with Time Windows

According to a problem in the real world, a mathematical model is established for the multi-depot vehicle routing problem with time windows (MDVRPTW). In order to improve the computational efficiency, first, based on decomposition and coordination technology (DCT), the problem is decomposed into several sub-problems and the customers are decomposed into coupling and non-coupling customers by a heuristic method. Second, due to the bad convergence when traditional DCT is used to solve the problem, the coordination values are designed elaborately by means of an adaptive genetic algorithm. Moreover, a tabu search method is developed to solve VRPTW for each sub-system. Finally, the validity of the proposed method is proved by the practical instance