The Optimization Of Multi-Depot Military Logistics Vehicle Scheduling Problem

With the development of economy and science and technology, many powerful military countries speed up the development of military logistics. And vehicle scheduling problem is a key ring in the military logistics system, which attracts more and more attention from domestic and foreign experts. However, the traditional military logistics vehicle scheduling problem is mainly based on artificial arrangement, and specified materials distribution by themselves depot, which has poor anti-strike capability, less consideration factor, low delivery efficiency and the higher operating costs. Therefore, in order to solve the deficiencies by a single depot vehicle scheduling in the actual situation of troops, The paper will study the optimization of multi-depot complications military logistics vehicle scheduling problems, which have some constraints situation about the level of competence distribution depot effects, the demand time windows of army, the loading capacity of the vehicle and request service of army, etc.. The main contents includes the following three aspects:Firstly, according to the several complex constraints, the paper established a mathematical model, which spent minimal transport costs of vehicle scheduling. After distribution capacity level for the depot is affected by many fuzzy factors, which leads to solving the more complex in the optimization of multi-depot military logistics vehicle scheduling problem, the paper proposed secondary fuzzy comprehensive judgment to carry out multi-factorial evaluation distribution of capabilities level of depot from the many fuzzy factors. Then, based on the assessed value of the size, the multiple depot problem turn into cycling field problem. It simplifies the complexity of the problems and improves the efficiency of solving problem.Secondly, when the simple genetic algorithm solve the vehicle scheduling problems, there are some shortcomings from poor local search ability and premature convergence, etc.. This paper improves the simple genetic algorithm by dual population strategy and predatory search strategy. Wherein, the dual population strategy can upset the balance within populations, through two populations independently evolved to exchange the outstanding individuals between populations. But, after running certain algebraic, two different populations may also appear similar situation of individual solutions. while the predatory search strategy can dynamically adjust crossover and mutation probability of groups, which make up the shortcomings of dual population strategy to reduce convergence rate. Combining two strategies not only enrich the diversity of the population to prevent premature convergence, but also strengthen the ability of local search to improve its quality of solutions.Finally, in order to demonstrate the improved effectiveness of the proposed algorithm in solving military logistics vehicle scheduling optimization problems in this paper, the improved algorithm is applied to an instance of military exercises by programming simulation. By compareing with the simple genetic algorithm and parallel genetic algorithm, the improved algorithm has better stability and higher solution quality in this paper; through compareing with the traditional vehicle scheduling method, the improved algorithm proposed could save costs and improve service quality in this paper.