Research On Bi-level Model And Method In Supply Chain Management

Fierce market competition forces enterprise mutual reciprocity with its suppliers and customers to meet user's request in such aspects as performance, style, quality, price, delivery date and service for the products. Thus forms a new supply chain management mode. How to build a conformable supply chain model to provide decision support with quantitative method is an interesting topic deserved study.Bi-level programming is widely used in recent years. It emphasizes global optimization of system that has hierarchical structure. That's just characterization of supply chain problem. So bi-level programming is used to modeling and solve supply chain problem in this dissertation. Bi-level model and method in supply chain management are researched based on summarizing relative theory in the paper, the main content is:1. Hierarchical structure of supply chain decision-making in the distributed management mode is analyzed. Bi-level model and method in supply and procurement decision-making are summarized and a procurement model is built. Logistics center location decision-making is analyzed from two angles: Combining customer needs and Combining routing, and illustrate with models.2. Hierarchical structure of supply chain based on core enterprise in the collaboration management mode is analyzed. Hierarchical coordination theory, its application in supply chain management, advantages and disadvantages of hierarchical coordination mechanisms are discussed. Hierarchical coordination based on bi-level programming is put forward.3. Distribution center location bi-level programming model with fuzzy parameters is built. Transform the model into fuzzy chance-constrained bi-level programming model using confidence level of max-demand and confidence level of max- capacity. Then based on possibility theory, the model is transformed into certain bi-level programming model which can be solved directly. Considering the characters of the model, a direct search method based on genetic algorithms is designed. An example is provided to check the feasibility of model and method.