Research On Large-Scale Instant Distribution Based On Clustering

Vehicle Routing Problem(VRP)is one of the key issues in the logistics distribution process.As the time plays an increasingly important role in logistics and economic activities and customers' awareness of a time-effective logistics distribution is becoming more intense,the research on fast-response,safe and reliable vehicle routing becomes more and more important,especially with the vigorous development and promotion of e-commerce,the frequency of immediate delivery of large-scale customer orders in the region has become more frequent and has gradually become an important issue in urban distribution research.So the research on such issues becomes more necessary.However,there are difficulties in applying the existing theory to this problem.According to related literature studies,the traditional precision algorithm and heuristic algorithm unaptly figure out this type of issue due to slow solution and low efficiency.This paper adopts a two-stage algorithm,first clustering and scheduling,reduce the problem solving scale and use faster running algorithm to achieve customer order distribution.Not only can it can provide faster,punctual,and efficient delivery services,but also help companies save costs and make rational use of vehicles and human resources.Therefore,research on this issue has practical significance and necessity.Based on the analysis of the research of Vehicle Routing Problem(VRP)and clustering theory,this paper combines the background of instant distribution industry and the problems of H company,and introduces the two-phase heuristic algorithm into the study of large-scale instant distribution problems: first,the order clustering process uses greedy clustering(nearest neighbor algorithm)and maximum-minimum distance clustering to partition all orders in the region,translating the large-scale VRP problem into a small-scale VRP problem,reducing computational complexity and improving the solution efficiency,at the same time,the two clustering methods are analyzed from two clustering effective indicators(CP/SP)to evaluate the effectiveness of clustering.In the second stage,the clustering process in the previous stage converts vehicleproblems into TSP problems.In order to quantify the impact of control time on customer satisfaction,the paper introduces time penalty coefficients and then establishes a vehicle scheduling model with time windows(TSPSTW)and use the genetic algorithm solving the model and replaces the traditional static operator with dynamic crossover and mutation operators,which makes the solution faster and results better,thus creating conditions for optimizing the time window problem quickly and effectively.The customer order clustering and distribution are achieved through MATLAB simulation.Taking example C401 as an example,the TSPSTW model is simulated and run,and the order distribution schedule is obtained,including the distribution route,total cost,total mileage,order schedule,etc.Hereby get the delivery plan for all orders.From the running time point,it takes only 7-8 minutes to complete the clustering and delivery of 501 orders,which further validates the high efficiency of the two-stage algorithm.Finally,the advantages and disadvantages of the two clustering schemes are compared with clustering indicators and distribution economic indicator,which provides reference for H company to select the appropriate scheduling plan.