Research On Distribution Network Planning Of Cold Chain Logistics Based On Community Detection

In recent years,with the continuous improvement of residents' consumption level and the popularity of e-commerce,well-known domestic e-commerce companies such as Alibaba,JD,Suning,etc.have entered the fresh industry.Thus,the cold chain logistics industry has entered a period of rapid development.However,in terminal distribution of the cold chain logistics,the unreasonable setting of distribution depots and the vehicle route planning can't guarantee the distribution limitation.At the same time,the cold chain food deterioration still occurs frequently,which make the improvement of customer satisfaction fail.Therefore,rational planning of cold chain logistics distribution network is an important research topic in cold chain distribution.Aiming at terminal distribution network planning of cold chain logistics,this dissertation introduces the complex networks to the distribution region partition,which led to the establishment of the community networks model.After utilizing community detection algorithm to divide the community structure of the model and finishing the customer clustering.This dissertation sets up a distribution network LRP model of cold chain logistics based on community structure and then designs an algorithm to solve the selection of distribution depots and distribution routing planning.This dissertation mainly completed the following work.(1)Converting communities with cold chain distribution demands in cities into nodes in complex networks,and the connection relationship between communities in the road network into edges in the network.Based on the factors such as distance between communities,road congestion,road grade and traffic light number,this dissertation comprehensively evaluated the travel time of cold-chain logistics distribution among communities and then,converted its reciprocal into the weight of the edge in the network.Thus,a community networks model was established,which make transform the problem of distribution region partition into find community structure of the community networks model.(2)Considering the capacity constraints of distribution depots in the distribution region partition,the capacity constraints were added to the search process of community structure.Louvain algorithm with capacity constraints was designed to divide the community network model and discover the community structure,thus customer clustering was completed.(3)Based on the community structure and considering the timeliness requirement in cold chain distribution,a multi-objective LRP optimization model is established with the minimum total network cost and the highest total customer satisfaction as optimization objectives.The multi-objective model was transformed into a single-objective model by setting a lower limit of total customer satisfaction.The corresponding genetic algorithm is designed to solve the problem and complete the selection of distribution depots and the distribution routing planning.(4)This dissertation makes a case study of SF company's cold chain logistics distribution network planning in Qingshan District.After establishing the community network model,the community structure is acquired by community detection algorithm to finish customer clustering.Based on the results of customer clustering,the selection of distribution depots and vehicle paths in each distribution area were solved and optimized by MATLAB software.Through comparative study,the meaning of counting cargo loss cost on LRP model solution results is verified.