| Summary: | 碩士 === 國立高雄第一科技大學 === 系統資訊與控制研究所 === 101 === In semiconductor manufacturing, the process of wafer fabrication is the most technologically complex and capital intensive stage. This large scale and discrete event process is highly reentrant with hundreds of machines and processing steps. It needs an efficient and effective scheduling method for large size reentrant flow shop problem in order to increase the competitiveness of the manufacturing company.
The reentrant flow shop problem (RFSP) means that a set of n jobs need to be processed on m machines following M1, M2, …, Mm and every job must be processed in certain machines more than once.
This research provides an effective Fuzzy-based Multilevel Genetic Algorithm to solving RFSP with the objective of minimizing the total turn around time (TTAT). The method that fuzzy-based multilevel genetic algorithm is set up a critical point on the basis of the scheduling problems characteristics and the objective function. According to the position of the critical point and current generation, fuzzy logic controller chooses the focused term of chromosome, then the genetic algorithm effects on this term.
Considering the large size of RFSP in this research, the objective function explicitly as TTAT. And TTAT = TAT1 + TAT2 + … +TATn, therefore our goal is to shorten the evolution process of each TAT. According to this proposed method, it is dynamically optimization and more efficiently retained better results. The fuzzy-based multilevel genetic algorithm is proposed to enhance the performance of original genetic algorithm.
In the end, the efficiency of the proposed approach is proved by computational experiments compared with former algorithm.
|
|---|
