Research Of Delivery Vehicle Routing Problem Based On Customer Segmentation

With the development and popularization of e-commerce, the logistics industry has developed rapidly, but the increasing number of the customer leads to a growingly larger scale of the logistics network and a low-level service efficiency and quality. In order to promote the development of e-commerce well, the existing logistics service must be improved. Distribution is a very important part of logistics links, and is the last link of the link from manufacturers to the consumers, but it is also the link with frequent delay, therefore the improvement in distribution efficiency can solve the logistics bottleneck effectively. How to select the appropriate vehicle routing which can quickly response to customer demand is the attention to the problem in this paper.Because of the large number of customer, it’s difficult to design the transport route directly, the calculation also is a trouble.So we can refer to the thought of two phase method, keep the vehicle routing problem divided into two stages, first reducing the network scale, that is, grouping the customers, and then on the basis of grouping for path planning. This method can effectively solve the mass customers vehicle routing problem.This paper studied on the vehicle routing problem based on above background, and aimed at the problem of the larger scale of customer and the objective function without comprehensive consideration, a method based on the customer group and the vehicle route planning solutions was put forward. In this method, firstly, select the customer clustering variables, build customer classification system, and express language variables with triangular fuzzy number; Secondly, evaluate of the attributes and their corresponding each customer comprehensively, calculate the relative importance weights of each attribute, cluster customers with fuzzy clustering method, then use the Dijkstra algorithm for path optimization for the grouped customer to find the shortest path in each group; Finally, with the lowest total transportation cost as the function target, build a VRP improved model. On the basis of the cost of transport routes, the objective function adds the penalty cost of the customer service time window and human cost(considering the importance of human resources). And use the Genetic Algorithm with implicit parallelism and global space searching characteristic to solve the model, getting the approximate optimal solution. This paper selected one day’s order data of a certain distribution center as the experimental data to verify this improved model, and the experimental results showed that the method presented in this paper achieved feasibility andpracticability on the problem of vehicle routing problem.