Solving Job-Shop Scheduling Problems by Boltzmann Genetic Algorithm

• Main Authors: Chi-Hsun Chung, 鍾奇勳
• Other Authors: Ruey-Huei Yeh
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/63294904306332949902

碩士 === 國立臺灣科技大學 === 工業管理系 === 97 === Job-shop scheduling problem (JSP), which was wildly used in industries, plays a vital role in manufacture scheduling. Many of the high-tech industries such as semiconductor industries, TFT-LCD industries belong to the Job-shop scheduling. Nevertheless, due to the variation of JSP, its combinatorial optimization problem in scheduling is recognized as one of the most complicated NP-hard problems. Many experts and scholars use the Generic algorithm to seek out the JSP problem, and its powerful searching ability of Genetic algorithm (GA) was widely applied in scheduling problem. However, the insufficiency of searching partial area in GA makes the process of evolutionary searching easily fall into the local optimal solution, lowering the efficiency of seeking out the optimal solution. Based on this phenomenon, this study combines GA with Boltzmann function in Simulated Annealing algorithm, which is characterized as not easily fall into local optimal solution, developing Boltzmann Genetic Algorithm (BGA), and aims to compare the quality and efficiency between BGA and traditional GA in minimum makespan in JSP. The result of this study indicates the advantageous of BGA over traditional GA in seeking out JSP, suggesting that the BGA can save extra time and cost, and benefit industries in planning manufacturing scheduling.