The Multi-Depot Vehicle Routing Problem With Simultaneous Deterministic Delivery And Stochastic Pickup

The vehicle routing problem is the core problem of logistics distribution enterprises.With the rapid development of economy,the distribution mode of joint distribution of multi-depot in a region is gradually formed.At the same time,the logistics distribution under the uncertain environment has become the difficulty of the development of logistics enterprises.Therefore,considering the multi-depot joint distribution,the return rule after the completion of vehicle service,the situation that customers have the demand for goods distribution and collection at the same time and the demand for goods collection is a random variable,the research on the vehicle routing problem with simultaneous deterministic delivery and stochastic pickup based on joint distribution can be more close to the actual needs.Reasonable planning of distribution vehicle driving path and dividing each path to the distribution center can effectively reduce the distribution cost of enterprises and improve the competitiveness of enterprises.At present,There are few researches on MDVRPSDDSP,most of the existing studies only considers single distribution,and the research on vehicle return rule is insufficient.Based on this,a two-stage MDVRPSDDSP model is constructed to solve the problem of the multi-depot vehicle routing problem with simultaneous deterministic delivery and stochastic pickup based on joint distribution(MDVRPSDDSP).In pre-optimization stage,the pre-optimization scheme is generated based on the stochastic chance constrained and vehicle capacity constraints.In re-optimization stage,we propose an improved re-dispatch policy to deal with the failure point in pre-optimization scheme.According to the characteristics of the problem,a metaheuristic combined adaptive genetic algorithm and variable neighborhood descent is designed to solve the problem.In this metaheuristic,aiming at the shortcomings of genetic algorithm such as fast convergence speed and weak local search capability,the deep search ability of the variable neighborhood descent algorithm is applied to the genetic algorithm to enhance the local search ability of the algorithm.The strategies include adaptive neighborhood search times strategy and adaptive mechanism of selective acceptance of inferior solutions are used to balance the diversification and exploitation in the algorithm iteration process.The validity of the model and algorithm is verified by several examples.Numerical results show that the model and metaheuristic we proposed is rather effective.The research results not only deepen and expand the VRP related theoretical research,but also provide a scientific and reasonable method for logistics enterprises to draw up the vehicle scheduling plan.