A guided-based genetic algorithm for solving assembly planning problems

• Main Authors: Chang Yin-Ho, 張銀和
• Other Authors: Tseng Hwai-En
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/56339236111759501569

碩士 === 大葉大學 === 工業工程學系碩士班 === 92 === Assembly planning refers to the task where planners arrange a specific assembly sequence according to the product design description as well as to their particular heuristics in putting together all the components of a product. Unlike traditional studies where the liaison graph goes with a genetic algorithm, an attempt is made to solve problems in assembly planning by using genetic algorithms (GAs) under the connector-based environment. The key point in this approach to assembly planning is to combine the connector concept and characteristics of a genetic algorithm using object-oriented programming. Because connectors serve as concept product building blocks in the design stage, more engineering features can be covered. With higher levels of information, the degree of complexity in assembly planning can be effectively reduced. Based upon the result of connector-based assembly planning model made by Tseng, Li and Chang (2004), the author proposed the so-called guide genetic algorithms to solve the massive constraints-oriented problems in assembly planning. The method of the traditional genetic algorithms is one kind of stochastic blind search procedure. Therefore, when the question constraints pattern is more complex, traditional GAs will create massive infeasible solutions. The procedure will reduce the GAs solution quality and the efficiency. In the mechanism of the guided GAs, initial population, crossover, mutation will be discuss in this research. Finally, a stapler, electric fan, and a laser printer were used as practical examples to demonstrate the possibility of such idea. Consequently, in terms of assembly planning, it is feasible to use the guided GAs to apply more complex product. This research also discovered that the efficiency of the guided GAs surpass the algorithms proposed by Tseng, Li and Chang (2004).