Research And Application On Large-scale Network Of Logistics Route Planning System

With the growth of Technology, e-commerce develops rapidly and the marketingmode of business to business becomes popular, which promotes the logistics industry.As we all know, decreasing the cost is very important for improving the logisticsefficiency. Nowadays, many scholars have studied on the Capacitated VehicleRouting Problem (CVRP) and given advices on algorithms, which are very successful.So this thesis is based on the logistics platform which involves the demand of severallogistics companies, and analyses three different modes of CVRP with generalities.The thesis mainly applies algorithm of Ant colony genetic fusion and the main partsare as follows:1. Based on the mode of CVRPTW(Capacitated Vehicle Routing Problem withTime Windows),we analyses the influence of multi-time windows on mode of CVRP.Each branch owns two time windows-morning and afternoon, which has influence onthe delivery routes.2. This thesis involves three modes of CVRP: one-vehicle and multi-path mode,multi-vehicle and multi-path mode, mixed path mode. It brings the mode of CVRPwith line templates, which results that the bound vehicles can only delivery amongthe branches in the line templates and the non-bound vehicles can only deliveryamong the branches out of the line templates, and it is called the mixed path mode.This thesis makes a detailed description and mathematic modeling of the three CVRPmodels.3. Combining algorithm of Ant colony genetic fusion with algorithm of geneticto solve three modes of CVRP. On the one hand, we improve algorithm of Ant colonygenetic fusion combining selective exploring and random exploring and the strategyof information updating combining global optimum strategy and iterative optimumstrategy; on the other hand, we improve algorithm of genetic combining two ways ofvariation, which enhances the ability of search,thus showing a better optimizationperformance.In order to validate the presented algorithms, we achieve the above algorithms, use the provided data to test the algorithms and contrast the improved algorithms indetail. Four experiments are done based on whether the time windows or theline-templates are used or not.This thesis analyzes the results of the experiments andthe convergence of the algorithm.The improved algorithm has been compared withsome other algorithms in every aspects, which proves that the improved method inthis thesis is reasonable and effective.