Study On The Fuzzy Optimization Design Of Bi-level Distribution Network In The Supply Chain

In customer-oriented manufacturing environment, driving force of enterprise has been changing from production to value added by distribution and service. Therefore, distribution network design is one of the most important ways to improve the customer satisfaction level and to strengthen competitive power of enterprise. Considered the uncertainty of customers' demands and production capacity of branches, the precise values of parameters are difficult to obtain. In this paper, the customers' uncertain demands are treated as fuzzy parameters and the problem becomes a fuzzy decision problem, which is more reasonable.In this paper, the fuzzy optimization design of bi-level distribution network under single-maunufacturer-mult-supplier-multi-distributor environment is studied. The fuzzy optimization design of bi-level distribution network includes model establishment, which is the important part in this paper, and its solution.First, it is how to establish a model. From the standpoint of integration and coordination of the supply chain, a fuzzy chance constrained programming model is presented for bi-level distribution network design;in which demand allocation is considered, under single-maunufacturer-mult-supplier-multi-distributor environment. The model embodies the uncertainty of customers' demands, non-shortage and customers' satisfaction of services. Because of exist of fuzzy parameters in aim function and subjected function, the model has no real mean. According to Zadeh's possibility theory and based on Liu and Iwamura's researches, the model can be converted into deterministic model by converting chance constraints to their respective crisp equivalents.It is difficult to solve model by traditional methods. Hence, genetic algorithm(GA) is used. The method of coding, the design of fitness function, initialization of population and design of genetic operators are discussed. The efficiency of GA to solve this problem is verified by a numerical example. Enterprise can make production plans and arrange for transportation according to results.The design can cut down the cost of distribution and production. And it's fit to those multi-product and small scale enterprises.