The Research On MDVRPTW Based On PSO With Mutation

Logistics distribution which plays a outstanding role in modern logistics is a very important part. On the one hand, the decrease of the cost of logistics distribution equals to the decrease of the total logistics cost, so it has become to be a important path to optimize the total logistics cost; on the other hand, it's the most intuitionist part for the customers' feeling in the whole logistics process, so the quality of logistics distribution service influences mostly the customers' evaluation, and it is one of the most important measure to improve customers' satisfaction. However, the condition logistics facing is become more and more complicated, just as the roadway network is perplexing, and the customers are located asymmetrically and so on. So, design reasonable and effective logistics distribution a scheme to decrease the distribution cost and improve the service level has become a practical key problem. The key problem we need to solve is how to use the vehicles effectively and adopt a most economical and reasonable routing path, which can be transformed to be a question deserved to research into: how to apply modem mathematics methods and computer on the problem.Using PSO algorithm is considered to be one of the most efficient ways to solve the NP Problem. In order to solve the MDVRPTW(Multi-Depot Vehicle Routing Problem with Time Window), a solution was proposed by combining Particle Swarm Optimization (PSO) with mutation operator. The mutation operator will change when the equation of adaptation changes. So we can break away from the local optimum, and strengthen the possibility of searching the global optimum. When applying the PSO algorithm, the initial solution is got by a specific method but not stochastic, so both the ability of search of optimized solution and the convergence speed can be improved. In addition, a new coding method which especially aimed at Multi-Depot Vehicle Routing Problem was put forward to reduce invalid iteration .Result of the experiments showed that the new algorithm not only has great advantage of convergence property over PSO, but also can avoid the premature convergence problem effectively.