半導體產業由於科技的進步而蓬勃發展,成為台灣最重要的製造業之 一。但由於全球半導體製造業生產的競爭激烈,企業的經營除了不斷創新 科技之外,同時要考慮如何提昇裝配電子產品的效率,以增加電子產品的 競爭力成了首要的目標。近年來,由於多元化的電子產品使得印刷電路板 (Printed Circuit Board;PCB)組裝製造傾向於少量多樣化生產。因 此,本研究針對此情況下研擬發展一套演算法,以求解表面黏著技術 (Surface Mount Technology;SMT)生產系統中組裝不同PCB類型之置料 槽架元件置換問題。 在多元化的PCB組裝製造中,會產生不同種類PCB的生產過程轉移,轉 移過程包含了元件的裝載、卸載、置換位置所需的時間,本研究考慮在兩 條生產線上處理多種類印刷電路板的製造組裝問題。本文中研究的目標是 考慮雙機之下所有不同種類PCB的加工順序,使得裝配完成時間為最短, 進而決定置料槽架上元件位置之置放問題的啟發式法則。本文中將對PCB 處理順序及元件置料槽架上設計演算法進行效果比較分析,以提供元件指 派問題上的應用。
In this thesis a production-planning problem arising in the assembly of printed circuit boards (PCBs) of various types and various volumes using two SMD placement machines with multi-slot feeders, is studied. The objective of this study is to determine the assembly sequences for board types and feeder rack assignments of the two machines so that the completion time of all requested orders is the minimum. Relatively few studies have dealt with cases of multiplicity of board types and multi-slot feeders. Adjustment to feeder setup is required each time a board type is changed during the manufacturing process. Two heuristics are proposed to solve the two machines dispatching problem. The first heuristic is P1-G2. It finds an initial solution for the case of one single SMD machine by the best nearest neighborhood method and then the solution is improved by 3-Opt and Or-opt procedures. The final solution is partitioned into two sequences on the equal time basis. The second one is G1-P2, it first determines the PCB types family for two machines and then finds good solutions for both PCB families. The solutions are then improved by randomly interchanges of PCB types in the two families and 2-Opt improvement procedure. Our experimental results show that heuristic G1-P2 outperforms heuristic P1-G2.