在過去單一廠區的自動化環境之下,廠內的生產進度的追蹤與控制,一般是由製造監控系統(manufacturing execution system; MES)來實現。近年來,隨著產業的快速擴張,在訂單不斷增加的情況之下,大部份的企業往往為了擴充自身產能而不斷地進行廠際間的合併與新廠的擴建,產成所謂的多廠製造環境。由於大多數的製造監控系統皆以單一廠區為設計時的主要考量,因此在面對多廠環境時,受限於各廠的成立時間、技術認知、成本限制、以及系統汰換上的差異,使得各廠因廠間資料的差異而無法有效地共享與交換彼此的生產資訊,發生多廠資料的整合問題。為了能在多廠環境之下監控在製品在不同廠的加工進度,使資料具備整體企業的觀點,製造監控系統必需具有資料收集以及整合的機制。 本研究的目的即針對MES在多廠之下因不同系統資料儲存裝置以及資料定義人員的不同所形成的差異化現象(稱之為資料異質化現象),提出一個有彈性、可適應異質環境的多廠資料整合架構(Multi-plant based data integration framework; MPDIF)。本研究在資料整合的過程裡使用了資料整合代理人、廠區代理人、以及資料擷取代理人這三個代理人模組來解決資料的異質化問題並整合各廠區MES之資料,同時達成多廠生產活動監視與追蹤的特性。 本研究所發展MPDIF整合模式,由於資料的擷取以及資料的提供皆為獨立的軟體代理人在運作,因此模式具備了整合複雜度低、術語名稱標準化、以及可應用在其它系統整合的優勢。本研究的貢獻不僅改良了傳統的資料整合架構,由於代理人獨立分層執行,因此可整合廠區間不同層次之異質資料,模式中也使用排程演算法進行代理人的派工,提升資料擷取代理人之執行效率。
A number of factories often use manufacturing execution system (MES) to monitor and control the processes of production in single-plant and environment. In recet years, many companies growth faster and the orders from customers are more and more. Every factories adopt some strategies such as factoies merge or build new factories in order to finish all orders. In either way, the factories will face a multi-plant production environment. There are many differences between factories such as start time of operation, technology, cost, and system change in multi-plant environment and it makes data exchange and data share impossible. Because of MES is work bettetr with the single-plant, there are not any functions to monitor the progress of production of orders in all factories. To be effective, MES must have some capabilities to collect and integrate data between different factories. This paper presents a system model for MES data integration, called multi-plant based data integration framework (MPDIF). The objective of the present model is to reduce all differences between factories we mentioned above in multi-plant environment. MPDIF uses factory agent, extraction agent, and integration agent to implement the capabilities of MES data integration in all factories. In addition, the integrated data is also used to trace all activities of prodution in every factory. In this research, the model we presented was not only have low complexity in integration, but also standardize all local terms in uniform way. In addition, MPDIF framework could support other information systems such as accounting system, ERP system, or PDM system. The model have three contributions: (1) improve the traditional model for data integration, (2) have ability to integrate data in different layers, and (3) use schedule rule to assign the routing path in data collection.