Study On Fuzzy Optimization Method For Global Production Planning Problems

Production planning problems study how to make appropriate production plans to get the maximum profit in the production process, and this is of great strategic significance for the long-term operation of enterprises. In the actual production process, demand, freight and exchange rate are usually uncertain, so this paper employ fuzzy theory to deal with the uncertainties in production planning problems.This thesis first establishes a fuzzy two-stage model with fixed possibility distributions, the goal of which is to maximize the profit. When the exchange rates are fuzzy variables, we propose a new two-stage fuzzy optimization method, and generate a measure by credibility distribution, which can define the L-S integral of the function of fuzzy variables. We analyze the characteristics of the model, discuss the properties of model solutions, and finally solve the model by using domain decomposition method. Considering the exact possibility distributions of the fuzzy variables in some decision-making problems are unable to determine, we further discuss a two-stage production planning model with variable possibility distributions. Using the value at risk reduction method of type-2 fuzzy theory, we obtain the parameters possibility distributions of fuzzy variables, and analyze the influence of parameter variation on the solution results. In the end, we provide some numerical experiments, and verify the influence of parameters on the solution results by comparing the results with the case of fixed possibility distributions.The major results of this thesis contain the following four aspects:(i) Two classes of two-stage fuzzy production planning models with fixed possibility distributions and variable possibility distributions are proposed;(ii) We discuss the basic properties of the models with general continuous possibility distributions, and derive some theoretical results;(iii) When the exchange rates are fuzzy variables, we propose a domain decomposition method to solve the models;(iv) We design the algorithms to solve the models, and provide numerical experiments under two cases to verify the influence of parameters on the solution results and demonstrate the superiority of parameter optimization method.