|
Leachman 針對半導體製造的產能利用，提出產能負荷模型 Capacitated
Loading Model），其透過時程參數（Cycle Time）來推導動態生產函數
（Dynamic Production Function）中的係數，並根據產品的需求，使用
線性規劃（Linear 有效分配工作站產能。然後，利用線性規劃軟體，解
出使公司獲利最大的投料計劃。雖然產能負荷模型功能完整且為最佳解，
但其缺點為需很長的電腦求解時間，數學規劃計算軟體昂貴，電腦主記憶
體需求極大，計算方法複雜，需有這方面專門訓練之工程師才能維護此套
系統。本論文的研究重點，即是保留產能負荷模型所提出的生產計劃假設
以及其動態生產函數，提出利用獲利率（Profit Ratio）為投料優先順序
之啟發式演算法取代線性規劃。利用此演算法代替線性規劃所求出的解，
乃是一趨近最佳解之近似解。以獲利率（Profit Ratio）為投料優先順序
之前提下，發展出三種啟發式演算法，分別為極大投料、單純平均投料及
成本平均投料啟發式演算法，其獲利率考慮了生產成本、存貨成本及欠貨
成本。最後分析比較以線性規劃及幾種啟發式演算法所求得生產投料計劃
之獲利的差異。結果發現，當所有產品皆需使用的瓶頸工作站愈少，啟發
式演算法之結果愈接近最佳解。另一方面，設計訂單即時詢問系統，考慮
訂單利潤是否合理，目前產能是否足以完成此訂單，可即時(Real Time)
決定訂單是否可接受；如產能不足，則可得知最早可交貨日期為何。
Leachman proposes a capacitated loading model to utilize the
factory efficiently. He uses cycle time to derive dynamiction
of a factory and use a linear programming (LP) formulation
toces among various products to optimize the corporate cash
flow. Although the model generates optimal solution for the
company, drawbacks: (a) it takes a long time to solve the LP
problem; (b) software is very expensive; (c) it needs a lot of
computer computation is very complicated; thus,(e) specially
trainedeeded to maintain this system. The main focus of this
research is to use the modelinghe Leachman''s capacitated
loading formulation and the method to dynamic production
functions; and to propose three heuristicn a profit ratio
calculation. These heuristic algorithms are totimal solutions.
In this research reject, we test the heuristics byeuristic
solutions and the optimal solutions under different problemed
on the experiments, the less number of shared bottleneck
machines the heuristics perform. In addition, we suggest a real-
time order quotation algorithm. proposed real-time quotation
system, which utilizes the sale plan the heuristic algorithms,
the salesmen can decide promptly if the promised.
|