Research On Vehicle Dynamic Allocation For Demand-driven City Distribution

With the pro-active implementation of logistics policy measures and the developmentof chain business and E-business, city distribution demand of multi-type, small batch, highfrequency, randomness, decentralization and time-effectiveness has been increasing. Itprovides not only a broad market space, but also a gigantic dynamic force for servicesuppliers. However, present distribution service level and social pressure (traffic jam,vehicle pollution, for example) cramp the development of city distribution. How to breakthrough the bottleneck and seize the opportunity is a key problem. In view of theimportance of transportation in city distribution operations, in addition, raising theutilization rate of vehicle can effectively reduce total vehicle number, relieve traffic jamsand protect the environment against automobile pollution, so this paper mainly studies theallocation of vehicle under analysis of the development trends of city distribution.Current conditions of study, researches on the model and algorithm of vehicleallocation are abundant, but on the combination of demand analysis and vehicle allocationare little. In addition, many studies are concentrated in method exploration and refinementof demand forecasting, but little on real-time demand data processing. In consideration ofthe above-mentioned facts, this paper proposes a detailed method for demand analysis andcombines it with vehicle allocation in order to provide theoretical reference for distributionservice suppliers.First, based on the definiteness of city distribution trends, this paper analyses theproblems of non-demand-driven vehicle allocation and static allocation, and puts forwardthe corresponding solutions, that is using the vehicle dynamic allocation model fordemand-driven city distribution. Then, from the standpoints of city distribution demandanalysis and vehicle dynamic allocation, this paper elaborates on the construction processof this model. Here, demand analysis includes order selection, order classification anddetermination of service priority, the adoptive methods are hard clustering and SOFMclustering. Vehicle dynamic allocation model aims at the whole lowest of the punishmentcost, vehicle operation cost and no-load ratio, with the loading capacity and volume as therestraints. Finally, the paper chooses44orders of A company at random as the sample dataof model, then uses Matlab and LINGO to get the results. The results indicate that themodel proposed in this paper is helpful to reduce distribution costs, shorten transportationtime and satisfy customers. Besides, the effective combination of order classification and vehicle allocation can improve arithmetic speed and thus assure the operation efficiency.