半導體產業顧客產能互換程序與排程變動演算法之研究

Abstract

本文針對半導體產業短期產能供需失調的情況，在多工單等級環境下，以不延遲已下單顧客交期為前提，面對客戶插單的要求，試圖協調產能之供需，以達到三贏的局面。 首先顧客產能互換程序共分為篩選模組、排程變動演算法、評估模組、與徵詢模組，其乃由既定排程、既有製程資料與新訂單資料放入篩選模組，找出潛在產能者，接著再進入排程變動演算法來排定優先順序，經過評估模組得出可替換訂單後進入徵詢模組進行詢價議價，最後使得顧客產能互換程序得以完成。 而生產排程變動演算法共分為負荷分布相似度與產品相似度。而以負荷分布相似度來看瓶頸機台原有產能利用情形，以及欲替代之新訂單使用瓶頸機台產能之情形。接著產品相似度的產生即是利用所推導之演算法來推估原有訂單與新訂單之相似程度。

Short-term unbalance of capacity between supply and demand happens frequently in semiconductor manufacturing. Under the condition that not delaying the confirmed orders, this thesis intends to solve the capacity exchange problem so as to transfer the capacity of a prescheduled order to the inserted order. First, customer capacity exchange procedure is formed by the screen module, algorithm for schedule change, evaluation module, and request for quotation module. Let the known schedule, process, and the information of new order get into the screen module, to find some candidates. Then let the data be the input to run the algorithm for schedule change, and there would be a priority. Through the evaluation module, we could have the changeable orders menu, and then we could request for the quotation and negotiate prices. After these processes, we could finish this procedure. Next, the algorithm for schedule change is formed by the capacity similarity between the given orders and the new order, and by the product similarity also between the given orders and the new order from the bottleneck machine. The capacity similarity shows both the capacity of original situation and also the changed situation, and let us realize which given order is suitable to exchange. And the products similarity is used by algorithm to formulate the degree of similarity between given orders and new orders.