Research On Data-Driven Dynamic Vehicle Routing Problems And Algorithms

With the rapid development of economic globalization and the continuous promotion of science and technology,modern logistics has been one of the most active and fast-growing industries in China.In 2019,the National Development and Reform Commission issued "Opinions on Promoting the High-quality Development of Logistics to Form a Strong Domestic Market",emphasizing the importance of fully utilizing new technology to develop smart logistics.As a key part in the logistics transportation system,the distribution problem plays a vital role.In the urban logistics distribution,the introduction of dynamic vehicle routing problem makes it closer to real life.While reducing costs,improving distribution operation efficiency and user satisfaction,it also eases traffic pressure.Therefore,the dynamic vehicle routing problem is an important part to realize intelligent logistics distribution,and it has been a research focus.In this study,the data of urban traffic checkpoints are analyzed,combined with data driving,and then the relevant traffic characteristics are extracted.A lot of relevant literature and research results are analyzed and summarized around dynamic vehicle routing problems.With the objectives of minimizing the cost of vehicles and transport time,a problem model is established,proposing an improved genetic algorithm suitable for it.With the simulation example of the SF logistics distribution,the scientific validity of the proposed algorithm is also verified.The main research works of this paper are as follows:(1)By summarizing the current research status and results of dynamic vehicle routing problems,this paper clarifies the significance of exploring dynamic vehicle routing problems on the basis of urban traffic data.From the theoretical results and practices of the traffic data mining and dynamic vehicle routing problems,this paper summarizes common types of traffic data and related mining algorithms,and outlines the vehicle routing problems and their classification.It also focuses on the relevant theories of dynamic vehicle routing problems,and elaborates the classification and solving algorithms of dynamic vehicle routing problems in different scenarios considering both domestic and foreign research results.(2)Regarding to the data in traffic checkpoint in Zibo City,this paper uses random forest algorithm to mine and classify traffic data by standardization,and abstracts the time-dependent speed segmentation function.This paper also establishes a mathematical model for the dynamic vehicle routing problem with time windows in a single depot,to simulate the logistics distribution problems in Zibo City.(3)Aiming at this problem model,this paper proposes an improved genetic algorithm for the dynamic model.The time updated function and adaptive strategy are introduced to optimize the vehicle dispatching scheme,in order to improve the distribution efficiency.Meanwhile,through simulation examples of actual distribution in SF Logistics,this paper compares and analyzes the optimization performance of the improved algorithm with and without adaptation,which further proved to be reasonable and effective in solving the dynamic vehicle routing problem.