The development of simulation-based heuristics in solving parallel machine scheduling problem

• Main Authors: Yu-An Shen, 沈育安
• Other Authors: TahoYang
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/33375438317264763464

碩士 === 國立成功大學 === 製造工程研究所碩博士班 === 93 === 　There are not only many machines installed in the work station of wire bonding of the semiconductor packing process, but also the bottleneck in this production line. The capabilities of the machines are different because of a lot of machines in the station, and times of purchased machines were different. Therefore, the wire bonding station is the classic uniform parallel machine problem. Schedule planning for parallel machines becomes the important factor of the improvement of production. In scheduling problem algorithms, the classic optimum schedule planning simplified the variables and conditions of process environment make the gap between model and real system. Furthermore, the simulation-based heuristic of intelligent search can not satisfied the requirement of in time for simulation. Therefore, this research will develop simulation-based heuristic algorithms to satisfy the requirement of immediately suggestion of scheduling. 　In this research, we develop three simulation-based scheduling heuristics, there were push algorithm, pull algorithm, and grouping machines of pull algorithm. We simulated different conditions of production, to observe the responses of different algorithms for objectives of service level, average utilization, flow time, average time in system, average queue time. Afterwards, this research used the TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) to estimate this five objectives and computed the weight by AHP (Analytic Hierarchy Process) to find the satisfied scheduling algorithm. The result showed that the representation of pull algorithm were better than other two algorithms, and became the proposed scheduling method in this research.