Order Assignment Problem Based On Diversified Demands

Since the value of fast-moving customer goods is very low, and the profit margin is very narrow, the fast-moving consumer goods industry is very sensitive to logistics costs. With the increasing number of the plants, the widely distributed and messy customer demands, and the increasing number of product items, it becomes very difficult for the enterprise to coordinate the production and delivery. To aim at this problem, this paper studies the order assignment problem with multiple plants, multiple demands and multiple items. The purpose is to find a reasonable order assignment way, in order to help the enterprise make effective production plan and distribution plan under limited resources, thus help the enterprise satisfy the customer needs and also get the biggest profit.The order assignment problem studied in this paper involves many complicated factors, according to the characteristics of demand, which is diversity and frequency, this paper first pre-treats the initial customer orders before the modeling. The order pre-treatment include three parts, they are time batch processing, customer location clustering and product items clustering. After the pre-treatment, the paper establish the order assignment model, the model includes capacity constraints and vehicle constraints with the objective of profit maximization.After modeling, the paper promotes two kinds of methods to solve the problem. The first are two exact algorithm, they are exhaust algorithm and branch-bound algorithm. It is proved that the exact algorithm can only solve the small problems or medium-small problems. Since the order assignment problem in the enterprise is usually of large scale, the variables of the model is huge, therefore, the paper puts forward a tabu search algorithm to solve large scale problems. Finally, this paper gives a simulation example to compare the tabu-search algorithm and the branch–bound algorithm, the result of the simulation show that the tabu search algorithm performs much better in computing time, and it also gets satisfactory solution.