應用基因演算法於印刷電路板之電容元件取置作業規畫

Abstract

在印刷電路板組裝(Printed Circuit Board Assembly, PCBA)製程中，影響組裝時間的主要原因為元件指派(Component assignment)與供料器指派(Feeder assignment)。目前中、外文獻中所針對的取置機皆屬於表面黏著機，此類型的取置機與並聯式三軸機器人(Delta Robot)在設計上有許多相異之處。因此，本研究希望藉由有效率地規劃Delta Robot的電容元件取置作業，使產品的生產週期時間越小越好。並利用基因演算法（Genetic Algorithm, GA）尋找在不同選擇方法與交配方法的組合下，求出較適當的電容元件指派及供料器指派，使得產品生產週期時間降低，並期望提供業界應用於實際生產環境中。 本研究利用台灣某知名之電腦主機板製造廠所提供之資料進行分析比較，探討在不同選擇與複製方式、交配方式及突變方式的組合下之差異。結果發現使用競賽法搭配PMX交配方式與雙點突變方式可以求得較低的生產週期時間。

In the printed circuit board assembly (PCBA) manufacturing process, component assignment and feeder assignment are the most important factor to impact the assembly time. Currently, all researches focus on surface mounted machine, which has many differences from Delta Robot in the design level. Accordingly, this research tries to minimize the production cycle time by planning the schedule of Delta Robot. In order to improve the schedule of Delta Robot, this research uses genetic algorithms (GA) to find the most appropriate component assignment and feeder assignment by comparing different selections, crossovers and mutations. This research uses the data from a well-known motherboard manufacturer to analyze the solutions within different selections, crossovers and mutations. The results show that the match of partially matched crossover, tournament selection and double mutations can acquire the lowest production cycle time.