根據德國工業4.0報告，未來智慧工廠於各個生產環節過程中需蒐集大量資料，並將其結合為一物聯網，以期機器設備皆具有獨立自主能力，自動完成生產線操作，而設備間透過互相溝通即時監控週遭環境，自動偵出問題並通知人員處理。為確保作業品質，產業間設備不斷地往自動化發展並加入大量感知器，除降低人員作業可能風險，並可隨時偵測機台狀態。隨著自動化程度愈高，異常停機會導致連續性機台停機，因此機台即時健康監控系統於自動化導入後是必需的，若未能於機台發生異常前提早預警並立即派人修復，可能會導致更嚴重品質風險。以半導體封裝產業設備，透過SECS/GEM讓主機與機台溝通，並要求機台內部資訊拋出，以將資料集中彙整分析。由於機台狀態特徵眾多，因此利用馬氏-田口系統(Mahalanobis-Taguchi System, MTS)篩選關鍵機台狀態特徵，刪除不必要及多餘的機台狀態特徵，以增加作業效率及降低分析成本；再合併計算移動變異數/共變異數矩陣之行列式值，並代入指數加權平均管制圖(EWMA)及換算機台健康分數，供工程人員判斷機台健康狀況之依據。本研究重點結論為：(1)減少33.33%機台狀態特徵數量，從24個降為15~16個，降低資料分析成本及增加處理效率；(2)利用移動變異數可以消除不同機台製造程式差異；(3)提供管理者及工程師較簡易的方式即時了解機台狀態。

As yield control becoming more and more rigorous, most industries expect to detect defective items prior to the completion of production. In this study, the main focus is to create a machine health index based on data collected by Semiconductor Equipment Communication Standard (SECS/GEM) such as temperature, pressure, voltage and others information to establish model to improve production quality.

However, collecting too much machine characteristics is always associated with high cost and low process efficiency. In addition, using too much useless machine characteristics may actually decrease the accuracy of prediction model. Thus, we select the critical machine characteristics using the MTS first to reduce the dimension of the problem. Then, we adapt a moving variance-covariance statistic to eliminate the difference in average responses between various recipes. The variance-covariance matrix is converted to a scalar by taking its determinant. The determinant of moving variance-covariance is then translated to a EWMA statistic and finally a heath score. This health score can be referred to as an index which can significantly reduce engineers’ time to monitor the all machines.

The proposed method is applied to wafer grinding process in semiconductor assembly and we found that: (1) The MTS is effectively in reduce the number of critical machine characteristics; in our case, a 33.33% reduction of the characteristics quantity from 24 to 15~16 which improves efficiency of data processing and also decreases the cost of analysis; (2) The moving variance-covariance statistic does eliminate the difference in average between recipes; (3) The machine health index is simple and easy to use.

英文部分
Bourke, P. D. (1991). Detecting a shift in fraction nonconforming using run-length control charts with 100% inspection. Journal of Quality Technology, 23(3), 225-238.
Chakhunashvili, A. & Bergman, B. (2006). An EWMA Solution to Detect Shifts in a Bernoulli Process in an Out‐of‐control Environment. Quality and Reliability Engineering International, 22(4), 419-428.
Chen, A. & Blue, J. (2009). Recipe-independent indicator for tool health diagnosis and predictive maintenance. IEEE Transactions on Semiconductor Manufacturing, 22(4), 522-535.
Chen, A. & Wu, G. (2007). Real-time health prognosis and dynamic preventive maintenance policy for equipment under aging Markovian deterioration. International Journal of Production Research, 45(15), 3351-3379.
Cudney, E. A., Hong, J., Jugulum, R., Paryani, K. & Ragsdell, K. M. (2007). An evaluation of Mahalanobis-Taguchi system and neural network for multivariate pattern recognition. Journal of Industrial and Systems Engineering, 1(2), 139-150.
Gan, F. (1990). Monitoring observations generated from a binomial distribution using modified exponentially weighted moving average control chart. Journal of Statistical Computation and Simulation, 37(1-2), 45-60.
Gan, F. (1993). An optimal design of CUSUM control charts for binomial counts. Journal of Applied Statistics, 20(4), 445-460.
Gan, F. (1994). Design of optimal exponential CUSUM control charts. Journal of Quality Technology, 26(2), 109-124.
Han, J., Pei, J., & Kamber, M. (2011). Data mining: concepts and techniques: Elsevier.
O’Sullivan, P., Martinzez, J., Durham, J. & Felker, S. (1996). Advanced fault detection for the semiconductor industry. Future Fab International, 1(1), 71.
Reynolds Jr, M. R. & Stoumbos, Z. G. (1999). A CUSUM chart for monitoring a proportion when inspecting continuously. Journal of Quality Technology, 31(1), 87.
Sandireddy, S. & Jiang, T. (2005). Advanced wafer thinning technologies to enable multichip packages. In Microelectronics and Electron Devices, 2005. WMED'05. 2005 IEEE Workshop on (pp. 24-27). IEEE.
SEMI. (2000). SEMI E5-1000 SEMI equipment communications standard 2 message content (SECS-II).
Su, C.-T., Wang, P.-C., Chen, Y.-C., & Chen, L.-F. (2012). Data mining techniques for assisting the diagnosis of pressure ulcer development in surgical patients. Journal of medical systems, 36(4), 2387-2399.
Su, C. T. & Hsiao, Y. H. (2007). An evaluation of the robustness of MTS for imbalanced data. IEEE Transactions on Knowledge and Data Engineering, 19(10), 1321-1332.
Su, Chao-Ton, et al. Data mining techniques for assisting the diagnosis of pressure ulcer development in surgical patients. Journal of medical systems 36.4 (2012): 2387-2399.
Taguchi, G., Chowdhury, S., & Wu, Y. (2005). Taguchi's quality engineering handbook. 2005. Jhon Wiley & Sons Inc.
Taguchi, G. & Jugulum, R. (2007). The Mahalanobis-Taguchi strategy: a pattern technology system.
Taguchi, G. & Rajesh, J. (2000). New trends in multivariate diagnosis. Sankhyā: The Indian Journal of Statistics, 62(B); 233-248.
Taguchi, G., S. C., Yuin Wu. (2004). Taguchi's Quality Engineering Handbook.
Wang, H. (2002). A survey of maintenance policies of deteriorating systems. European Journal of Operational Research, 139(3), 469-489.
Wu, C.-T. & Kuo, J.-Y. (2007). Utilization of Bayesian Reliability Theory on the Investigation of IC Handler Machine Preventive Maintenance Timing in Semiconductor Testing House. Journal of Data Analysis, 2(3), 123-138.
Xiao, L., Song, S., Chen, X. & Coit, D. W. (2016). Joint optimization of production scheduling and machine group preventive maintenance. Reliability Engineering & System Safety, 146, 68-78.
Xu, F.-L., Zhao, Z.-Y., Zhan, W., Zhao, S.-S., Dawson, R. & Tao, S. (2005). An ecosystem health index methodology (EHIM) for lake ecosystem health assessment. Ecological Modelling, 188(2), 327-339.
Yang, T. & Cheng, Y.-T. (2010). The use of Mahalanobis–Taguchi System to improve flip-chip bumping height inspection efficiency. Microelectronics Reliability, 50(3), 407-414.

中文部分
李知珈 (2013)，「半導體廠機台即時製程狀態變異偵測值之分析」，成功大學工業與資訊管理學系學位論文。
徐祥禎 (2009)，電子構裝結構分析，Retrieved from http://stam.org.tw/newsletter2a.php?SN=15&opp=download1
張皓然(2012)，「發展支援向量資料描述-田口系統 (SVDD-TS) 於多變量分類問題之應用」，朝陽科技大學工業工程與管理系學位論文。
彭茂榮 (2016)，2016年第二季台灣IC產業回顧與展望，IEK產業情報網。
游欽宏 (2005)，半導體與平面顯示器製造自動化技術，機械月刊，31(12)，6-23。
廖勝泰 (2000)，SECS/GEM/HSMS Practice. Retrieved from 財團法人工業技術研究院 機械工業研究所。
趙安國 (2008)，「批次剖面資料之錯誤偵測分析」，清華大學統計學研究所學位論文。
蕭宇翔、蘇朝墩和沙永傑 (2004)，「應用 MTS 於非平衡資料分析之穩健性研究-以行動電話檢測流程為例」，國立交通大學工業工程與管理學系學位論文。
蘇朝墩 (1997)，產品穩健設計 田口品質工程方法的介紹和應用，中華民國品質學會。
蘇朝墩 (2002)，品質工程，中華民國品質學會。

網站部分
DISCO. (2014a). Dry polish wheels: gettering DP. Retrieved Dec. 15, 2016, from https://www.disco.co.jp/cn_t/products/catalog/pdf/dgp8761.pdf
DISCO. (2014b). Fully automatic grinder/polisher DGP8761. Retrieved Dec. 15, 2016, from https://www.disco.co.jp/cn_t/products/catalog/pdf/getteringdp.pdf