Multi-type Vehicle Routing Problem With Time Windows Based On Hybrid Evolutionary Algorithm

Vehicle routing problem(VRP)has important practical significance and is highly complex.It is reflected as the core problem of logistics management in our real life.Its applications include goods distribution in supermarkets,industrial garbage collection,school bus route,which are closely related to our life.In multi-type vehicle routing problem with time windows(MTVRPTW)is more in line with the current research,which not only meets the special time window requirements of customers,but also facilitates the distribution company to adopt diversified distribution methods and realize the reasonable allocation of distribution resources.Due to the NP-hard characteristics of MTVRPTW,and in order to overcome the disadvantages of a single algorithm,the hybrid evolutionary algorithm with a mixture of multiple algorithms or strategies is an effective way to solve the problem,which has positive theoretical and practical research value.In this paper,a hybrid evolutionary algorithm is proposed to solve multi-type vehicle routing problem with time windows.Firstly,the mathematical model of VRPTW is constructed,and reasonable encoding and decoding is designed,and effective genetic operator is set to complete chromosome gene recombination.The hybrid evolutionary algorithm includes fast sampling strategy-based global search(FSS-GS)and route sequence difference-based local search(RSD-LS).FSS-GS can realize fast convergence to Pareto front surface and RSD-LS is used to further improve the search ability of the algorithm.A series of experiments on Solomon benchmark problem show that the proposed method and strategy have better performance in convergence and adequate distribution performance.Secondly,the mathematical model of the multi-type VRPTW is improved.The individuals were expressed in the form of two chromosomes.A reasonable crossover operator and mutation operator were designed to complete the gene recombination of two chromosomes.On the basis of the proposed hybrid evolutionary algorithm,the hybrid algorithm is divided into several stages,and the use time of RSD-LS is adjusted.The experimental results show that using appropriate optimization strategy at different stages of the algorithm can not only save the calculation time of the algorithm,but also further improve the solving ability of the algorithm.Compared with the optimization algorithms such as NSGA-?,SPEA2 and MOEA/D,the proposed method is more excellent in solving the MTVRPTW.The proposed hybrid evolutionary algorithm in this study,combines the global search strategy based on fast sampling strategy and local search strategy based on issues related to well balancing algorithm search and exploring ability.At the same time,the algorithm is divided into several different stages,and the precise use of optimization strategy can further improve the effectiveness and efficiency of the algorithm.The methods and strategies proposed in this study provide useful reference for hybrid evolutionary algorithms to solve complex multi-objective optimization problems...