| Summary: | 碩士 === 國立成功大學 === 工業與資訊管理學系 === 103 === Multi-level linear programming (MLLP) is an approach that can be used to solve decision problems based on hierarchical decision structures. Satisfactions that represent the judgement of solutions obtained for a decision-maker (DM) include uncertain information. Hence, the concept of fuzzy sets is used to define the satisfaction of the corresponding value by building membership functions. This study develops a method for interpreting the relationships of satisfaction between different levels by using satisfaction intervals. To overcome the shortcomings of methods that strictly restricts the satisfaction of lower-level DMs by higher-level DMs, a more flexible decision space is proposed. The proposed model implies that satisfaction obtained of lower-level transcending the one of higher-level is allowed. An exceeding ratio (α) is proposed to provide an exceeding tolerance for lower-level DMs under some limited conditions. Moreover, linguistic variables are used to express the extra requirements of DMs that restrict the relation between adjacent levels. After integrating all constraints, fuzzy goal programming is applied. A comparison of the proposed method with other fuzzy goal programming approaches in solving multi-level programming problems shows that the proposed method achieves higher satisfaction when the middle level is difficult to satisfy. This study provides a method for interpreting the satisfaction between different levels by giving a more flexible decision space for lower-level DMs to reach higher satisfaction.
|
|---|
