The Development of Flow Shop Scheduling System Using Genetic Algorithm

• Main Authors: Kai-wei Liu, 柳楷韋
• Other Authors: Jyh-bin Suen
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/67944932464467995531

碩士 === 義守大學 === 工業工程與管理學系碩士班 === 96 === Scheduling is a type of resource allocation decision-making solution. Under limited resources, it can effectively manage and allocate the production sequence for each task, so as to achieve the optimal overall performance of its production line. Flow shop scheduling problem belongs to NP problem, and in recent years, evolutionary algorithm has been used to solve these type of compositionality problem. While this research integrated genetic algorithm and Matlab calculation software together, to develop a visualized flow shop scheduling teaching system, with the purpose of discussing the minimum makespan and the hope of finding the optimal work schedule. Besides, it enabled teachers to use the scheduling solution in relevant course teaching, through user interface operation and visualization display. Aiming at the solution of the genetic algorithm, four kinds of Crossover and two kinds of Mutation were adopted and were supported by the elite policy, in order to prevent the evolutionary retrogress, to accelerate the constringency speed of the genetic algorithm, and to find a better solution. After running it through the systematic test, the research compared it with GA-Wang and GA-Lian, and then measured it with mean error rate, to discuss its solution performance. The result indicated that, Order Crossover and Order-based Crossover paired up with Exchange Mutation and Shift Mutation created the best constringency effect, followed by the Position Crossover paired up with Exchange Mutation and Shift Mutation, with the second best effect. Furthermore, the poorest effect was generated by the Cycle Crossover paired up with Exchange Mutation and Shift Mutation.