Study On Optimization Of Location Routing Problem Considering Customer's Pickup And Home Delivery

In recent years,with prosperity of information technology and e-commerce in China,lifestyles of consumer and shopping style are more and more diversified.While e-commerce drives the rapid development of logistics from online shopping,it also promotes the explosive growth of express package volume,the cost of the "last mile" distribution is high in the logistics industry.The express delivery industry is facing the challenges of scattered customers,diversified demand from consumers and stronger consumer awareness of rights protection.At present,the common problem for e-commerce and logistics enterprises is how to reduce the cost of terminal distribution by establishing a perfect urban distribution system on the premise of ensuring customer satisfaction.Based on the project from National Science and Technology Plan named "technology research and application demonstration on logistics under online shopping and city joint distribution service ",this paper continues to study the location routing problem of terminal points around joint distribution,and to study the research and application of collaborative cooperation from the view of e-commerce platform and logistics enterprises.For e-commerce platform,using the massive historical order data stored in it,through data exploration,feature engineering,this paper finally uses machine learning of GBDT algorithm to analyze and model the historical order data,so as to realize the prediction of future goods distribution volume and guide the subsequent path planning.In view of the logistics system,based on the consideration of two distribution modes of customer's delivery and home delivery,by introducing the corresponding distribution utility function,under the limited rational decision-making mode,this paper constructs the customer's delivery utility function with distance as the core,and constructs the home delivery utility function with comprehensive consideration of cost and time window.This paper set up the multi-objective location routing model with the lowest comprehensive cost and the largest customer distribution utility function,the scanning algorithm is designed to improve the non dominated sorting genetic algorithm(NSGA-II),and the scanning algorithm is used to generate the initial population,which is finally verified by the model data solution.On the basis of the above model,the location path planning problem with time window is further considered,a vehicle path planning model comprehensivly considering cost and the time penalty cost is constructed.Under the condition of considering the vehicle load and the time window at the same time,the insertion algorithm is used to improve the initial population of the genetic algorithm.At the same time,in order to retain the best information of the parent,a complete path in the chromosome is selected as the crossover operator.Finally,the synthetic model data is used to verify the problem.