本計劃之目的在於提升中衛體系中之管理效率並電子化中心廠與衛星廠間之聯繫。中 心廠面對為數眾多的衛星廠，如何為每一張訂單選擇最佳之外包供應鍊，即形成所謂 的動態供應鍊組成問題。本計劃擬透過軟體代理人來自動搜尋合適的衛星廠，其蒐搜 尋機制乃建立於最佳化模型上。本計劃為三年期。第一年計劃擬利用前期計畫所建立 之代理人平台引進拍賣機制，其中，中心廠的角色為訂單拍賣主，而衛星廠則為投標 者。衛星廠邊以模糊多目標規劃考慮本身之產能限制與可能訂單排程以決定其標單內 容；而中心廠則以多屬性決策方法決定最佳標單。第二年計劃進一步在決策過程中加 入協商機制並以多層分散規劃來描述中心廠與衛星廠之間的層級關係。中心廠可藉其 訂單決策影響衛星廠之報價與排程決策；另一方面，衛星廠亦可藉由其提出之報價與 排程決策改變中心廠之訂單決策；此一過程體現了協商的精神。第三年計劃著重在當 系統發現外包商之產能產生非預計之變化時，可以針對新的產能限制重新規劃外包商 間之生產批量。在及時化生產的考量下，當代理人發現有外包商產能因設備故障等問 題而產能縮減時，必須立即通知中心廠與相關製程之外包商，並重新同步各單位間之 生產批量以符合拉式生產之精神。我們將以個案公司為例，在本計劃建立之代理人平 台上實作所提之各項決策機制，並以電腦模擬結果驗證本計劃所提方法之效率。 The purpose of this project is to promote the management efficiency and to automate the communication between firms in a center-satellite factory system. In the center-satellite factory system, the selection of appropriate satellite factories for each customer order becomes a dynamic supply chain formation problem. This project attempts to automate the search of appropriate satellite factories by employing software agents, whose decision-making mechanism are based on optimization models. The project spans three years. In the first year, we will introduce the auction mechanism into the agent platform established in our previous project, where the center factory plays the role of auctioneer and satellite factories bid for the orders. Satellite factories will use a fuzzy multiple objective decision making model to determine their bids, while the center factory chooses the best bid by employing a multiple attribute decision making approach. The second year project further introduces the negotiation mechanism into the decision making process. The hierarchical and interdependent relationship between the center and satellites factories is modeled by the multi-level decentralized programming. In the model, the center factory can guide the satellite factories』 decision in quotation and scheduling by its order decision. On the other hand, satellite factories can also change the center factory』s decision by their proposals. The third year project will focus on the synchronization of production lot size in the supply chain under the consideration of just-in-time production. When unexpected capacity loss is found in some satellite factories in the supply chain, the production lot size will be recalculated according to this capacity change. The dynamic synchronization of lot size can reduce the waiting time and hence avoid late delivery of the customer order. This project will conduct computer simulation on the agent platform, with the case company as a practical example, to evaluate the performance of the proposed approaches.