The Logistics Network Optimization Of Urban Express On The Clustering Algorithm With Constraints

With the rapid development of electronic commerce,express,as a special form of logistics service,is facing the explosive growth of business volume.As a result,it doomed that courier companies will be in expansion and improvement in the near future.Compared to other logistics services,express delivery requires not warehousing functions,but faster transport,lower cost and short time.Logistics network planning is the key to the service efficiency and quality of express enterprises,and the network layout has gradually become the most important area of competition between enterprises.According to city express logistics network optimization problem,this paper designed three-layer network topology including collection centers,transportation centers and distribution outlets,qualitative analyzing the function and operation ability of nodes in each layer and location method.In the process of customer clustering,it is proposed that the method of distance measurement should consider obstacle constraint like river and high speed road to meet the real situation.Moreover,an algorithm for computing obstacle distance in the large data environment was designed.In the process of express network building,the workload upper bound was set for each cluster to represent the trade off in operation.The problem of distribution area division is abstracted and modeled based on constrained clustering.The feasibility of this model is verified by the case analysis based on actual distribute location data in Shanghai.Integrating advanced research achievements and methods,this paper designed the method of distribution area dividing for urban express network based on data mining.The practical improvement of K-means clustering is carried out from the following three aspects.Firstly,the method of clustering distance is redefined based on the obstacle constraint.Secondly,the constraint of workload is set to control the clustering result.Last but not least,the algorithm can converge more quickly than the original clustering center by improve of select method of initial cluster center.After that,based on the constrained customer clustering algorithm,102501 customer data items produced by in Shanghai by an express company within a day was clustered into 113 distribution area,and the results are displayed in the map.