Research Of The Bidirectional Logistics Routing Problem With Time Windows Based On Hybrid Genetic Algorithm

The development of logistic industry has improved the social economy. VehicleRouting Problem (VRP), the key method to optimize delivering and transportationproblem in logistic industry, refers to a combinatorial optimization to organize thereasonable lane line and then minimize the cost of distribution for a customer groupwhich has a clear and open demand under some limited constrains. The research onVRP problems could help the logistical companies to lower the cost, consume theresource in a rational way, improve the efficiency, and provide the better service forcustomers. Currently, the market value of reserve logistics in the world is growingrapidly year by year. The Forward Logistics problems and the Reserve Logisticsproblems were often conducted separately in previous research on VRP research.However, in reality, companies and customers would combine these two problems inorder to lower the cost or protect the environment.This article focused on vehicle routing problem with simultaneous deliveries andpickups with time windows (VRP-SDPTW) based on Hybrid Genetic Algorithm. Firstof all, the author analyzed special problems about the VRP-SDPTW and thenestablished a mathematical model having certain versatility according to basicconditions of VRP-SDPTW. This model could be also applied to other vehicle routingproblems by setting constrains. On the basis of these applications, the author designedImproved Genetic Algorithm for VRP-SDPTW and conducted checking computationfor this new algorithm by using MATLAB and then got the conclusion that GeneticAlgorithm could solve VRP-SDPTW.Secondly, the author analyzed and compared processes and characteristics ofHybrid Genetic Algorithm and Simulated Annealing (SA) Algorithm, and thenproposed Hybrid Genetic Simulated Annealing (GSA) Algorithm used to solveproblems of VRP-SDPTW. In GSA Algorithm, the author applied the Metropolisrules into crossover steps of Genetic Algorithm and then introduced parentcompetition to compensate corresponding disadvantages of Genetic Algorithm bytaking the advantages of local research ability of SA Algorithm. The author achievedthe GSA Algorithm with MATLAB programming through the calculation example and then compared performances of GSA Algorithm with the performance of GeneticAlgorithm in third chapter. After analyzing the results, the author got the conclusionthat the GSA Algorithm designed in this article has validity, universality, goodresearch ability, efficient computing power and stability of the performance of thealgorithm.