| Summary: | 碩士 === 東海大學 === 工業工程與經營資訊學系 === 101 === Today for solving NP-hard scheduling problem, there are many outstanding achievements. In the past, many researchers mainly focused on proposing different heuristic algorithms, but studies of flexible job-shop scheduling based on mathematical planning were rarely addressed. This paper develops mathematical programming model according to Ozguven et al.’s work in 2009 (FJSP-PPFs) for dealing with job-shop scheduling problems that resemble-real-world production problems. Because of consuming much computing time when solving NP-hard problems, this paper solves problems by parallel computing. It divides the mathematical model into a lot of combinations (small problems), and then the divided combinations are assigned to the parallel computers to increase the efficiency of solving the problem. The study is carried out in three steps. In the first step, the small scale mathematical model is developed to illustrate the accuracy of the schedule planning results as its optimal solution is easily obtained. In the second step, the medium scale mathematical model is proposed to illustrate our schedule planning results are the same as those of FJSP-PPFs developed by Ozguven et al. In the third step, the large-scale mathematical model is developed to analyze the planning results and the efficiency of parallel computing. The experimental results show that it decreases 90% of planning time in the medium scale problem by parallel computing with 32 processors associated with the accelerated solving mechanism. It decreases 70% of planning time in the large scale problem by parallel computing with 64 processors combined with the accelerated solving mechanism.
|
|---|
