Research On The Vehicle Scheduling Problem In Military Material Distribution

With the development of military science and technology, the traditional mode of material support in PLA can not suit the requirement of modern war. In the light of the experience of armies in developed countries, we should explore the"distribution"mode for military material support. The vehicle scheduling problem is crucial for military material distribution, and is important to the optimization of the military logistics system. After reviewing related literature about the vehicle scheduling problem and some usual algorithms for vehicle scheduling, the vehicle routing problem and the vehicle routing problem with time windows of military material distribution were studied in this paper. The main contributions are given as follows:1) The literature about VRP, VRPTW, and military logistics distribution problem were reviewed, and the exact algorithms, classical heuristics, and meta-heuristics for vehicle scheduling problem were summarized.2) The mathematical model of the vehicle routing problem of military material distribution was formulated, in which the minimization of the latest delivery time was used as the objective, and the ant colony optimization algorithm was umtilized to solve it. In the proposed algorithm, three routing construction methods, including serial algorithm, parallel algorithm, and improved parallel algorithm, were designed. In addition, the Relocation, Exchange, and 2-opt heuristics were used to optimize the routes that is constructed by ants. The experimental results showed that the parallel algorithm outperformed the serial algorithm in precision, and the method we used to improve the parallel algorithm was successful.3) The mathematical model of the vehicle routing problem with time windows of military material distribution was established, in which the minimization of the total waiting time of all customers was used as the objective, and the improved ant colony optimization algorithm was used to solve it. In our algorithm, the transition rule was improved according to the character of the problem, and three local search algorithms were integrated into the ant colony optimizaiton algorithm. The VRPTW benchmark instances were solved under different parameter settings, and the appropriate values for paramters in the transition rule were determined in the light of experimental results.