Research On Differential Evolution Algorithm For Allied Vehicle Routing Problems With Multiple Transport Networks

Allied vehicle routing problem, under the framework of logistics union transport scheduling optimization, are hot topics related to the research of applied mathematics, network analysis, graph theory, computer applications and transport. In this thesis, the author carrys out mathematical modeling, differential evolution algorithm design, its improvement and model simulation based on the characteristics of AVRP with multiple transport networks, and the main contributions are as the following:1. Research on single AVRP with multiple transport networks:contraposed the specific characteristics of such problem, established the mathematical model of the problem, designed the effective differential evolution algorithm to solve it, at last it proved feasibility and effectivity using differential evolution algorithm to solve such problems.2. Research on AVRP with hard time window and multiple transport networks: carrying in vehicle routing problem, usually there is the extend characteristic with time windows, this thesis designs the corresponding mathematical model and the differential evolution algorithm, this algorithm adds the characteristic of dual population, avoid losing the diversity of single population, insure searching the best value in a large scope, so as to improve the ability finding the best value in the whole optimum.3. Research on AVRP with soft time windows and multiple transport networks:with the soft time windows every transportation task needs to be completed within the time scope, or punishment will be gived. In this thesis, contraposed the specific characteristics with the soft time windows, the differential evolution algorithm adds optimization strategy of niching, simulation achieves good results.4. Research on AVRP with transfer points and multiple transport networks:it's complex in practical vehicle routing, in every transfer points it will increase the cost of time and manpower; in this thesis, contraposing such condition establish the mathematics model and the simulation experiment, via the improved self-adapt differential evolution algorithm, it has a perfect result for saving money to solve the best tools combination.