隨著科技產業的蓬勃發展,積體電路的需求倍增影響著整條半導體生產鏈,積體電路的設計越來越多樣化,其後段製程封裝技術也隨之提升,在全球封測廠激烈競爭下,如何縮短交期及提高生產率是未來發展的關鍵。 本研究探討單廠生產訂單排程及平行機台指派問題,在半導體晶圓封裝製造上,產量的大小取決於瓶頸站銲線站機台使用率,而影響機台使用率取決於更改機生產程序的次數,隨著改機次數變多,可生產時間便隨之下降,進而降低產能,本研究以銲線站為排程目標站,找出最佳投單順序進而降低改機次數以達到機台使用率最大化。 本研究考慮的生產排程與平行機台指派問題為NP-Hard問題,我們首先提出了混整數規劃數學模式,並建立最佳化程式,將訂單資訊自動轉換成決策訂單排程所需資訊,透過Matlab軟體中的基因演算法函數來進行求解,當得到近似最佳解時,最佳化程式自動推算出晶圓研磨至銲線站之各站每日投料量,做為每張訂單在機台加工順序排程決策的參考。
Through the improvement of science and technology industry, the demand of IC increases and effects the whole semi-conductor process lane. The design of IC has become more and more diversification and the packing technic has also improved since the great compete between IC testing and packing company, decrease the due date and increase process rate is an important task. This research investigate on order scheduling and parallel machine assignment problem in single factory, in semiconductor wafer packing process, the out decide the usage of the bottleneck wiring and what effects the usage is the frequency of changing machine progress, once the frequency more often, the less its able to process and effect the output. This research focuses on wiring as target to find the best input order in order to decrease the frequency to the maximum machine usage. What we consider : scheduling and parallel-machine assignment is a NP-hard problem, first we propose mathematical model of mixed integer programming using auto transform order information into what is needed and use genetic algorithm in Matlab to solve, once we get similar optimal solution, the optimal program will automatically calculate the daily usage from wafer milling to wiring. This makes the consultation of the order of arranging.