薄膜電晶體顯示器TFT-LCD為我國重大的「兩兆雙星」產業。因機台及產品特性的關係，生產現場經常是24小時不停線，屬於高成本、高技術的高科技製造廠。電腦整合製造（Computer Integrated Manufacturing, CIM）系統主要是利用電腦控制網路與管理技術，將產品由工廠投入到成品產出的所有生產過程與控制整合在一起。由於掌控了生產現場機台及生產資料的紀錄，系統穩定度與工廠的產能更顯息息相關。而CIM系統因需整合各項子系統，而非常龐大複雜，且系統功能也持續不斷地變動。對於站在一線面對使用者解決異常的人員而言，要在最短時間內判斷異常原因，及提出解決方案及找到問題處理窗口，以儘快讓生產現場回復運作是件困難的工作。只能靠長期的值班經驗累積和員工訓練，來判斷問題的分類並推斷出適當解決窗口。而在人員流動時，由於經驗不足而拉長了系統異常的回復時間，也間接了影響了工廠既定的產出時間，甚至是交貨時間。目前個案公司在CIM系統異常處理輔助解決的部分，僅有過去的異常處理紀錄及參考文件可使用。本研究主要利用個案分析法，透過個案公司的專家訪談及技術文獻的整理分析，針對異常處理各階段的判斷邏輯進行歸納，研究並建置一套可供CIM客服中心使用的決策模式。研究結果發現，以人、事、時、地、物五大異常確認要素為判斷基礎的決策模式，透過個案公司實際發生的系統異常為測試案例驗證後，可輔助值班人員進行釐清問題，並能依循決策模式裡的推論模組，找到異常的解決方案。TFT-LCD is one of the most important industries in Taiwan. Due to the characteristics of its equipment and product, the production site operates 24 hours per day. It is both capital and technology intensive.A CIM system utilizes computer-controlled network and management technology. It integrates and controls the production process from the input to output in the factory. System reliability and capacity are more closely related to the control of equipment and the generation of production records on the production site. However, a CIM system can become very large and complex because of the integration of various subsystems, and system is continuously modified. It is very difficult for the help desk agents to determine the causes of production abnormalities, provide an effective solution, and find the system contact personnel to recover the production in a short timeframe. The agents usually identify system problems and dispatch contact personnel based on their experience and training without technology support. A change of agent then may extend recovery time, extending the scheduled lead time and the time of delivery.For supporting to handle the abnormality of the CIM system, the company case only has records how the event is processed and reference document. The main purpose is to improve the system continuously to prevent the recurrence of production interruptions. This research attempted to build up a decision model for the CIM Help Desk of company case with the case study method by interviewing domain experts and analyzing technical documents to generalize the determine logics each step on handling abnormality. This research found that the model base on the five factor of abnormality check: “People”, “Event”, “Time” ,”Place”, “Object”, which has been verified by system abnormality occurred in the company case, can really support the agents of the CIM help desk to determine and resolve system abnormal events, thus improving the production efficiency of the company.
