| The paper’s research target is Incident Vehicle Routing Problem (IVRP), in which exists certain connection between customer’s need. IVRP Optimization is widely applicated in production and circulation; it belongs to NP-hard problem. In this paper, the basic types of the IVRP have been studied by using the Electromagnetism-like algorithm. According to the different types of VRP, different mathematics models were built; examples are used to verify the feasibility and efficiency of the new algorithm. The main work of the paper is described as follows:Firstly, the paper researched a single vehicle routing problem, then built mathematical model, designed an improved EM-like algorithm. According to the discrete characteristics of transportation scheduling problem, a method was proposed to the solving the population particle encoding, and here adds an initial solution rule for produce first generation. The EM algorithm in chapter2adOpts a revised total force calculation, help to avoid the local Optimization.Secondly, mathematical model for multiple vehicles scheduling problems was built. Combined with the positioning method and2-Opt method on the routes for the partial improvement, it expand the search space for best solution, avoid the EM algorithm falling into local Optimum. Inspired by the nature of survival in the wild world, elimination mechanism was introduced. This mechanism eliminates poor particles in each iteration, and introduces better particles to the current population.Thirdly, in chapter4, the Incident Vehicle Routing Problem with time windows was discussed, and a new method of moving particles was produced.Fourthly, a parallel EM-like algroithm is put forward to solve VRP. A master-slave scheme parallel particle population algorithm based on coarse granularity is conformed. The algorithm adOpts "master/slave" synchronous paradigm, andAt last, contents of the paper are summed up and achievements are concluded, and the following plans and prospects are pointed out. |
PDF Full Text Request