| Summary: | 碩士 === 國立交通大學 === 資訊管理研究所 === 97 === Since the introduction of the flowshop scheduling problem by Johnson in 1954, scheduling in flowshops has drawn much research attention. This thesis considers scheduling in flowshops with supportive precedence relations. The model originates from a real production context of a chemical factory that produces foam products. We reconstruct traditional two-machine model to divide all operations into two categories: supportive items and major jobs. Supportive items and regular jobs are to be processed by the stage-one machine and the stage-two machine, respectively. In the model, many different compositions of foam can be mixed in the foam blowing stage (on machine one), and products are processed in the manufacturing stage (on machine two). Each job on machine two cannot start until its supportive perations on machine one are all finished and machine two is not occupied. The objective is to minimize the total job completion time. The studied problem is strongly NP-hard. In this thesis, we propose a branch and bound algorithm equipped with a lower bound and two dominance rules. We also design a heuristic and a meta-heuristic (ILS) to derive approximate solutions. The branch and bound algorithm can solve instances with up to 40 jobs and 10 items. From the statistics of computational experiments, it seems very efficient and effective for large-scale problems with up to 200 by the ILS approach. And the average deviation is less than 0.109% from optimal values in small-scale problems.
|
|---|
