Research On Time-dependent Logistics Distribution Based On Customer Clustering

In the current competitive business environment, logistics is linked closely to the market, the simple logistics and distribution services can not meet the requirements of enterprises. Especially, for the rapid development of the machinery, vehicles and other industries related, it has put forward higher requirements to the relevant logistics and distribution industry, and quick response to the needs of different customers and providing differentiated services to customers have become a basic requirement to the logistics and distribution services.In order to improve the efficiency of distribution services and customer’s satisfaction, a time-dependent logistics distribution modeling based on customer clustering is proposed. The modeling includes two parts, one is customer clustering, the other is goods distribution. At the first part, customers are divided and clustered according to different attributes. At second part, based on customer clustering, each customer group is provided with differentiated services according to the properties of different customer groups.First, for the target of quick response to the needs of different customers, a hybrid fuzzy-hierarchy clustering method is proposed, and customer demands are described into several relevant attributes. Customers are clustered according to these different attributes and provide differentiated services to each customer group according to the different needs of customers. So it improves customer satisfaction, reaches the target of quick response to the needs of different customers, and provides a basis for further distribution of goods.Second, for the requirement of reducing the supply side distribution costs and improving demand side customer satisfaction, the delivery is divided into two sub-stages: vehicles assignment and vehicle routing planning. In these two sub-stages, considering the target of the supply side and demand side, multi-objective function is built based on supply-side and demand-side. In the sub-stage of the vehicles assignment, supply-side enterprise can decide the order of the customer service group according to the actual strategy; In the sub-stage of the vehicle routing planning, supply-side enterprise can select the routing, according to the actual needs and preferences of the deciding-maker, in order to reduce en-routing vehicle running costs and improve customer satisfaction. Third, in the traditional distribution modeling, en-routing vehicles costs is considered as a fixed value. In order to overcome this problem, we consider the real-time traffic conditions, and the road of real-time traffic is divided into a number of high and low peak hours, all en-routing vehicles’speed is a linear function of time running. So, en-routing vehicles’costs is relevant with the time of the vehicle run into the road, and avoiding the sudden increase and decrease of the vehicles, making the planned path of the vehicle distribution more realistic.Fourth, for the number of exponential increase in the modeling of VRP with the size increasing, we use ant colony algorithm to solve this mathematical modeling, with the positive feedback, distributed computation and constructive greedy heuristic search feature, the ant colony algorithm can solve the multi-dimensional dynamic combinatorial optimization problem, and reduce the computation time.Finally, an example is proposed to analyze that the original distribution strategy of the enterprise is significantly improved with the proposed distribution mathematic- cal modeling in this thesis, and this modeling is also proved with the better practical value.