| Summary: | 碩士 === 元智大學 === 工業工程與管理學系 === 95 === With the changing of production types, taking setup time into consideration is getting more and more important for industrial schedulers. Therefore, in this research, we explore the single machine scheduling problem that all jobs have a common due date that has been predetermined using the median of the set of sequenced, the objective is to find an optimal sequence with minimizing the total absolute deviation（Tardiness& Earliness）and the unrelated parallel machine scheduling with minimizing makespan under Sequence-dependent Setup time.
Dominance Properties（DP） are developed according to the sequence swapping of two jobs and getting a new better sequence . Because DP trap into local optima easily, we combined dominance properties with global search algorithm, Genetic Algorithm and Simulated Annealing, became GADP and SADP. The hybrid algorithm is a two-stage procedure; the first phase result of DP is the initial solution of the second phase. We compared GADP and SADP with SGA and SA, the result shows that there is significant difference between them in single machine scheduling problems and Parallel Machine Scheduling Problems. In addition, DP outperforms GA and SA significantly in parallel machine Scheduling Problem. Therefore, DP is able to combine with meta-heuristic and get better solutions.
|
|---|
