Vehicle Scheduling Problem In Logistics System And Algorithm

At present, logistics not only become another profit source of enterprise beyond the production area, but also the core power of gaining the market competitive. The vehicle routing problem (VRP) is an important part which directly connected to consumers, and whose schedule is reasonable or not will affect the enterprise economic efficiency. Nowdays, there are still not reasonable transportant allocation and scheduling line because of lacking wright logistics technology, result in wasting resource and poor service quality. Therefore, there is great practical significance to study this project.In this paper, a few typical vehicle routing problems are researched:VRP, VRPTW and MDVRPTW. Three problems are the NP-hard which develop simple to complex. Intelligent optimization has become mainstream solution to solve these problems, such as Genetic Algorithm and Particle Swarm Optimization. With the deep research on algorithm, people find the basic algorithm shortcoming like that GA easy to mature convergence and PSO is not suitable for solving discrete combinatorial optimization. So, academics made improvement or algorithm fusion, and achieved good results. But it increases the complexity; or change the solutiong in searching process to fit the discrete modle, to a certain it broke the algorithm structure. So, according to the predecessor's study, this paper propose a new particle collision PSO which use traditional PSO principle, take GA operational means to determine the partical next move to solve problem in paper. Work of paper as follows:1, Introduce the meaning and purpose of study on VRP in detail, describe, define and classificate the problem.2, Make an overiew about solving algorithm, focuse on the principle and characters of GA and PSO, analys the difference between two algorithms.3, Establish basic VRP mathematic model, design implement GA and PSO. Based on this propose partical collision PSO and give mathematic theory to support. Use three algorithms to solve two instances, than analys and contrast results; it proved that the new algorithm is effective.4, Establish VRPTW mathematic model, and generate initial group based on Solomon VRPTW instances, sovle the C, R and RC. For the unbalance issues we can use the balance factor to adjust.5, Establish MDVRPTW mathematic model, analys the problem solution—step method and whole method. Design implements two methods to solve MDVRPTW instance. From the results there is no too much influence to algorithm by selecting which one.Finally, conclusions and further research directions are given.