應用基因演算法建構印刷電路板人工插件生產線排程系統

Abstract

印刷電路板組裝（Printed Circuit Board Assembly, PCBA）生產線上有許多的組裝工作，影響生產線上最主要的組裝工作是元件指派（Component Assignment）並使得生產線平衡（Assembly Line Balance），亦即如何有效的分配元件至各工作站。目前大多數與PCBA的相關研究是以組裝表面黏著式（Surface Mount Device, SMD）元件的表面黏著機為研究對象，極少專門探討雙列直插式封裝（Dual In-line Package, DIP）元件的組裝。由於DIP元件的尺寸與形狀不一，一般皆以人工方式插件，而各元件指派也必須遵循某些置放規則以保組裝品質。本研究以國內某知名之電腦主機板製造商之PCBA的人工插件生產線為研究對象，應用基因演算法（Genetic Algorithm, GA）為其建構PCB人工插件生產線排程系統，將人工插件的元件置放規則納入本研究所提之演算法中，以期能快速產生出生產週期時間短且生產線平衡率高的元件指派。本研究並將所得之元件指派成果與使用模擬退火法與禁忌搜尋法所得之結果相比，發現本研究所提出之基因演算法能夠得到較佳的元件指派。

There are many assembly operations in the assembly line for printed circuit board assembly (PCBA), and the most important operation affecting the assembly line is component assignment. To make assembly line balance, the key factor is how to effectively assign components to every workstation. Currently, most of the research objects of the related research about PCBA are surface mount device; few studies use a dual in-line package for printed circuit board as the research object, and the amount of a dual in-line package for printed circuit board depends on the engineers’ experiences to every workstation, which spends a lot of time. Being aim to the dual in-line package for printed circuit board, this study uses genetic algorithm(GA) to construct the procession system. We use different selections, crossovers and mutations to compare mutually, and the results show that the match of partially matched crossover, tournament selection and double mutations is the best, which makes the appropriate component assignment in a short period and shortens the cycle time of the dual in-line package for printed circuit board.