A Genetic Algorithm for Job Scheduling on Parallel Machine

• Main Authors: Hsin-Chiang Huang, 黃信強
• Other Authors: Chi-Chang Jou
• Format: Others
• Language: zh-TW
• Published: 2000
• Online Access: http://ndltd.ncl.edu.tw/handle/96291080791312524760

碩士 === 淡江大學 === 資訊管理學系 === 88 === In order to make the solution of manufacturing scheduling more realistic, we consider earliness/tardiness cost, machine idle time cost, and machine setup cost in the selection of sub-optimal solution. In the case of parallel-machine scheduling, the problem has been proven to be NP-hard. Several applications of Genetic Algorithms to solve the optimization problems have been proposed recently. They have been shown to obtain better results than other algorithms. Thus, this research will investigate a new type of chromosomes and define its evolution processes in a Genetic Algorithm for solving parallel-machine scheduling problems under the above mentioned factors. In this paper, we first propose a new mode of chromosomes to make search space of the problem more complete. Secondly, we redesign evolution processes and fitness functions to help the speed of obtaining sub-optimal solution more rapidly. Finally, we design and implement a working scheduling system and compare its results with those of previous studies. The experimental results are analyzed and we demonstrated our algorithm has the following advantages: 1. The proposed fitness function and chromosomes perform better than traditional genetic algorithms. 2. The obtained solutions are more stable in the same that the deviations are smaller. 3. It considers more factors, and is more suitable for real manufacturing cases.