Research On Vehicle Routing Problem With Time Windows Based On Two-stage Heuristic Algorithm

With the rapid development of e-commerce in the new era,the status of logistics and distribution has also become higher and higher,and has become an important cornerstone for the development of e-commerce.At the same time,its asymmetry with e-commerce development will also have an impact on e-commerce.To restrict.The core problem of modern logistics is optimizing asset allocation and improving service perception while providing the best service at the lowest cost.Therefore,planning the vehicle route in the distribution process and formulating a reasonable plan is the focus of vehicle routing research issues,especially the study of Vehicle Routing Problem With Time Windows.VRPTW can not only increase the output value of unit assets,but also optimize the customer service experience from the time dimension.From the perspective of logistics companies,the improvement of asset efficiency can reduce hardware investment,and the company's capital flow pressure will be greatly reduced,thereby improving the competitiveness of the entire industry and the treatment of employees,creating a virtuous circle.Therefore,the research on VRPTW is of great significance.Based on the above research significance of VRPTW,this paper uses VRPTW-related theories and algorithm theories to describe concepts and factors.Comparing various algorithms that can solve path optimization problems,and tests the heuristic algorithms by citing specific data.For comparison,the genetic algorithm was selected as one of the algorithms applied in this paper.Based on the vehicle constraints,the time constraints required by the customer,and the constraints of various relevant factors,a mathematical model was established with the goal of minimizing the cost of the transportation project.Combined with the initial stage of distribution in real life,chaos may occur.In this case,the K-means ++ algorithm was selected as the firststage algorithm,and the basic genetic algorithm was improved as the second-stage algorithm.Finally,six kinds of data in the Solomon dataset were selected for testing and compared with the currently published optimal results.It was verified that the model and algorithm established in this paper can be used as a reference for solving practical problems in the distribution process.