虛實整合系統是工業4.0的關鍵應用之一，是一種結合電腦運算以及感測器和致動器裝置的整合系統。虛實整合系統實現人與網路的整合計算。導入虛實整合系統會提高產品的質量以及附加價值，而其複雜程度與使用性也會相對提高，虛實整合系統本身會受到它的複雜程度及使用性上的限制，因此本研究嘗試建立一套以機器學習為基礎的虛實整合系統，降低其複雜程度並提高使用性。本研究利用機器學習手法在虛實整合系統的虛擬子系統裡建立一套錯誤偵測與分類系統，此系統分為兩步驟，首先利用隨機森林演算法去分析從虛實整合系統感應器收集來的數據，來確定其關鍵變量，讓工程師可以知道是那些因素造成機台故障問題。第二步是建立長短期記憶神經網絡的時間序列學習模型，實現對機器的實時監控。在機器發生故障之前，錯誤偵測將發出警報並允許工程師調整機器參數或安排預測性維護以減輕機器停止運作的影響。在第一個案例研究裡，所提出的方法優於其他時間序列技術，並且能前3小時預測機台故障，其準確率達到80%或更高，並且在二個案例研究裡改善281%的產品壽命，此方法也可應用於其他的複雜系統，以提高機器利用率和生產力。

Cyber Physical System (CPS) is one of the key application of industry 4.0. CPS is an integrated system that combines computing, sensors and actuators. CPS is controlled by computer-based algorithms that integrate people and internet. However, the performance of CPS is limited by its computational complexity. How to implement CPS with less computational complexity and implement diagnostics, forecasting and equipment health management more efficiently in a real time performance remains an important issue. Therefore, the study attempts to establish an integrative machine learning method to reduce computational complexity and improve applicability as a virtual subsystem in CPS environment. This study consists of two steps. First, Random Forest (RF) is utilized to figure out key factors and reduce the data dimension for less computational complexity such that process engineers can figure out which factors are critical to the failure of machines. The second step is to establish a time series deep learning model based on Long Short-Term Memory (LSTM) network to achieve real time monitoring of machines. Before a machine fails, fault detection will alarm and allow the engineers to adjust parameters of machines or arrange predictive maintenance to mitigate the impact of machine shut-down. According to two empirical studies, the proposed method outperforms other times series techniques and the accuracy achieves 80 percent or higher 3 hours before machine down in a real time manner in the first case and improve the 281% productivity of a process in the second cases. This method can be applied to other complex systems to boost the efficiency of machine utilization and productivity.

第壹章 緒論 7
1.1 研究背景與動機 7
1.2研究目的 7
1.3研究架構 8
第貳章 文獻探討 9
2.1 虛實整合系統（Cyber Physical Systems） 9
2.2 錯誤偵測與分類系統（Fault Detection and classification） 11
2.2.1 多變量分析 13
2.2.2 機器學習 16
2.2.3 預測性維護 17
2.3 文獻總結 19
第參章 研究方法 20
3.1 利用隨機森林進行特徵篩選 21
3.2 時間序列模型進行製程監控 23
3.2.1 時間序列 23
3.2.2 長短期記憶模型 24
3.2.3 評估績效 27
第肆章 案例研究 28
4.1 案例研究一 28
4.1.1 隨機森林的關鍵因數識別 29
4.1.2長短記憶神經網路建立與評估 31
4.2 案例研究二 36
4.2.1 導入虛實整合系統與改善績效 37
4.3討論 44
第伍章 結論與未來展望 45

圖目錄
圖1. 方法流程圖 20
圖2. 隨機森林示意圖 21
圖3. 長短期記憶神經網路結構（Gers, 2015） 25
圖4. 晶圓切割機台 29
圖5. 長短記憶神經網路均方誤差反覆運算圖 32
圖6. 機台每十分鐘之錯誤偵測圖 34
圖7. 機台之完整錯誤偵測圖 35
圖8. 破網的類型分布 37
圖9(a). 太陽能網版印刷機 38
圖9(b). 太陽能網版印刷機之感測器分布圖 38
圖10. 長短記憶經網路預測太陽能網版壽命的趨勢 41
圖11. 無線溫溼度感測器數據資料 42
圖12. 虛實整合系統硬體架構圖 42
圖13. 網版壽命統計 43

 
表目錄
表1. 時間序列模型比較 17
表2. 變量重要程度表 30
表3. 線性回歸、支持向量回歸和長短記憶神經網路之績效 32
表4. 提前預測準確度 35
表5. 重要因子表 39
表6 線性回歸、支持向量回歸和長短記憶神經網路的績效 40

Abdallah, I., Dertimanis, V., Mylonas, H., Tatsis, K., Chatzi, E., Dervilis, N., Worden, K., & Maguire, E. (2018). Fault diagnosis of wind turbine structures using decision tree learning algorithms with big data. Safety and Reliability–Safe Societies in a Changing World, 3053-3061.
Akkaya, I., Derler, P., Emoto, S., & Lee, E. A. (2016). Systems engineering for industrial cyber–physical systems using aspects. Proceedings of the IEEE, 104(5), 997-1012.
Baba, N., & Kozaki, M. (1992). An intelligent forecasting system of stock price using neural networks. In Neural Networks, 1992. IJCNN., International Joint Conference on. IEEE, 371-377.
Botezatu, M. M., Giurgiu, I., Bogojeska, J., & Wiesmann, D. (2016). Predicting disk replacement towards reliable data centers. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. ACM, 39-48.
Brockwell, P. J., Davis, R. A., & Calder, M. V. (2002). Introduction to time series and forecasting, 2.
Bursteinas, B., & Long, J. A. (2000). Transforming supervised classifiers for feature extraction. In Tools with Artificial Intelligence, 2000. ICTAI 2000. Proceedings. 12th IEEE International Conference, 274-280.
Calderón V, S. A., & Nieto, F. H. (2017). Bayesian analysis of multivariate threshold autoregressive models with missing data. Communications in Statistics-Theory and Methods, 46(1), 296-318.
Chai, T., & Draxler, R. R. (2014). Root mean square error (RMSE) or mean absolute error (MAE)–Arguments against avoiding RMSE in the literature. Geoscientific Model Development, 7(3), 1247-1250.
Chiu, Y. C., Cheng, F. T., & Huang, H. C. (2017). Developing a factory-wide intelligent predictive maintenance system based on Industry 4.0. Journal of the Chinese Institute of Engineers, 40(7), 562-571.
De Carolis, A., Tavola, G., & Taisch, M. (2016). Cyber physical systems in manufacturing: Future trends and research priorities. In 21st Summer School Francesco Turco 2016, 12-17.
Derler, P., Lee, E. A., & Vincentelli, A. S.（2012）. Modeling cyber–physical systems. Proceedings of the IEEE, 100（1）, 13-28.
Díaz-Uriarte, R., & De Andres, S. A. (2006). Gene selection and classification of microarray data using random forest. BMC bioinformatics, 7(1), 3.
Genuer, R., Poggi, J. M., & Tuleau-Malot, C. (2010). Variable selection using random forests. Pattern Recognition Letters, 31(14), 2225-2236.
Gers F., Cummins F., Fernandez S., Bayer J., Wierstra D., Togelius J., Gomez F., Gagliolo M., & Graves.A. (2015). Colah’s blog. Retrieved from http://colah.github.io/posts/2015-08-Understanding-LSTMs/.
Gombé, B. O., Mérou, G. G., Breschi, K., Guyennet, H., Friedt, J. M., Felea, V., & Medjaher, K. (2019). A SAW wireless sensor network platform for industrial predictive maintenance. Journal of Intelligent Manufacturing, 30(4), 1617-1628.
Górecki, T., & Smaga, Ł. (2017). Multivariate analysis of variance for functional data. Journal of Applied Statistics, 44(12), 2172-2189.
He, Q. P., & Wang, J. (2008). Principal component based k-nearest-neighbor rule for semiconductor process fault detection. In American Control Conference, 2008. IEEE, 1606-1611.
Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural computation, 9(8), 1735-1780.
Jiang, J. R. (2018). An improved Cyber physical systems architecture for Industry 4.0 smart factories. Advances in Mechanical Engineering, 10(6), 1687814018784192.
Jiang, Y., & Yin, S. (2018). Recursive Total Principle Component Regression Based Fault Detection and Its Application to Vehicular Cyber physical systems. IEEE Transactions on Industrial Informatics, 14(4), 1415-1423.
Johnson, R. A., & Wichern, D. W. (2004). Multivariate analysis. Encyclopedia of Statistical Sciences, 8.
Khair, U., Fahmi, H., Al Hakim, S., & Rahim, R. (2017). Forecasting Error Calculation with Mean Absolute Deviation and Mean Absolute Percentage Error. In Journal of Physics: Conference Series. 930(1), 012002.
Kim, D., Han, S. C., Lin, Y., Kang, B. H., & Lee, S. (2018). RDR-based knowledge based system to the failure detection in industrial cyber physical systems. Knowledge-Based Systems, 150, 1-13.
Kopetz, H. (2011). Real-time systems: design principles for distributed embedded applications. Springer Science & Business Media.
Krotofil, I., Derler, P., Emoto, S., & Lee, E. A. (2016). Systems engineering for industrial cyber–physical systems using aspects. Proceedings of the IEEE, 104(5), 997-1012.
Krotofil, M., & Cárdenas, A. A. (2013). Resilience of process control systems to cyber-physical attacks. In Nordic Conference on Secure IT Systems. Springer, Berlin, Heidelberg, 166-182.
Lee, J., Bagheri, B., & Kao, H. A. (2015). A Cyber physical systems architecture for industry 4.0-based manufacturing systems. Manufacturing Letters, 3, 18-23.
Leinweber, D. J. (2007). Stupid data miner tricks: overfitting the S&P 500. Journal of Investing, 16(1), 15.
Li, Z., Wang, K., & He, Y. (2016). Industry 4.0: potentials for predictive maintenance. In International workshop of advanced manufacturing and automation, Manchester.
Lin, J., Su, L., Yan, Y., Sheng, G., Xie, D., & Jiang, X. (2018). Prediction Method for Power Transformer Running State Based on LSTM_DBN Network. Energies, 11(7), 1880.
Liu, Y., Hu, Y., Zhou, R., & Wen, J. (2016). An approach based on improved grey model for predicting maintenance time of IPS2. Procedia Cirp, 47, 204-209.
Loh, W. Y. （2011）. Classification and regression trees. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery, 1（1）, 14-23.
Lu, A. T. H., Austin, E., Bonner, A., Huang, H. H., & Cantor, R. M. (2014). Applications of machine learning and data mining methods to detect associations of rare and common variants with complex traits. Genetic epidemiology, 38(S1), S81-S85.
Majaj, N. J., & Pelli, D. G. (2018). Deep learning—Using machine learning to study biological vision. Journal of vision, 18(13), 2-2.
Muñoz-Romero, S., Gómez-Verdejo, V., & Parrado-Hernández, E. (2017). A novel framework for parsimonious multivariate analysis. Pattern Recognition, 71, 173-186.
Nikolakis, N., Alexopoulos, K., Xanthakis, E., & Chryssolouris, G. (2019). The digital twin implementation for linking the virtual representation of human-based production tasks to their physical counterpart in the factory-floor. International Journal of Computer Integrated Manufacturing, 32(1), 1-12.
Nikolakis, N., Maratos, V., & Makris, S. (2019). A cyber physical system (CPS) approach for safe human-robot collaboration in a shared workplace. Robotics and Computer-Integrated Manufacturing, 56, 233-243.
Niu, X., Liu, G., Huo, L., Zhang, J., Bai, M., Peng, Y., & Zhang, Z. （2018）. Risk stratification based on components of the complete blood count in patients with acute coronary syndrome: A classification and regression tree analysis. Scientific reports, 8（1）, 2838
Pal, M. (2006). Support vector machine‐based feature selection for land cover classification: a case study with DAIS hyperspectral data. International Journal of Remote Sensing, 27(14), 2877-2894.
Rivas, A., Fraile, J. M., Chamoso, P., González-Briones, A., Sittón, I., & Corchado, J. M. (2019). A Predictive Maintenance Model Using Recurrent Neural Networks. In International Workshop on Soft Computing Models in Industrial and Environmental Applications, 261-270.
Selcuk, S. (2017). Predictive maintenance, its implementation and latest trends. Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, 231(9), 1670-1679.
Srivastava, S., Engelhardt, B. E., & Dunson, D. B. (2017). Expandable factor analysis. Biometrika, 104(3), 649-663.
Stanisavljevic, D., & Spitzer, M. (2016). A Review of Related Work on Machine Learning in Semiconductor Manufacturing and Assembly Lines. In SAMI@ iKNOW.
Strobl, C., Boulesteix, A. L., Zeileis, A., & Hothorn, T. (2007). Bias in random forest variable importance measures: Illustrations, sources and a solution. BMC bioinformatics, 8(1), 25.
Swischuk, R. C., & Allaire, D. L. (2019). A machine learning approach to aircraft sensor error detection and correction. Journal of Computing and Information Science in Engineering, 19(4), doi: 10.1115/1.4043567.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Boston, MA: Allyn and Bacon.
Tan, Z. X., Goel, A., Nguyen, T. S., & Ong, D. C. (2019). A multimodal LSTM for predicting listener empathic responses over time. In 2019 14th IEEE International Conference on Automatic Face & Gesture Recognition, 1-4.
Wang, Y., Shen, Y., Mao, S., Chen, X., & Zou, H. (2018). LASSO & LSTM integrated temporal model for short-term solar intensity forecasting. IEEE Internet of Things Journal, 6(2), 2933 – 2944.
Wu, H. C., Chang, C. T., Chen, B. H., Lee, C. Y., Chang, C. J., Ko, J., Zhou, M & Liang, M. (2003). Fault detection and classification of plasma CVD tool. In Proceedings of the 2003 IEEE international symposium on semiconductor manufacturing, 123-12.
Lv, M., Yu, W., Lv, Y., Cao, J., & Huang, W. (2019). An integral sliding mode observer for CPS cyber security attack detection. Chaos: An Interdisciplinary Journal of Nonlinear Science, 29(4), 043120.
Wu, W. F., Ou, K. L., Chou, C. P., & Hsu, J. L. (2003). PECVD Ti/TiN x Barrier with Multilayered Amorphous Structure and High Thermal Stability for Copper Metallization. Electrochemical and solid-state letters, 6(2), G27-G29.
Xu, L. D., & Duan, L. (2019). Big data for cyber physical systems in industry 4.0: A survey. Enterprise Information Systems, 13(2), 148-169.
Yan, K., Zhong, C., Ji, Z., & Huang, J. (2018). Semi-supervised learning for early detection and diagnosis of various air handling unit faults. Energy and Buildings, 181, 75-83.
Yuan, Q., & Wei, S. (2018). Aligning Network Traffic for Serial Consistency and Anomalies with A Customized LSTM Model. In 2018 IEEE International Conference on Progress in Informatics and Computing, 322-326.
Yuanyuan, L., & Jiang, S. (2015). Research on equipment predictive maintenance strategy based on big data technology. In Intelligent Transportation, Big Data and Smart City (ICITBS), 2015 International Conference on IEEE, 641-644.
Zhang, Y., Guo, Z., Lv, J., & Liu, Y. (2018). A Framework for Smart Production-Logistics Systems based on CPS and Industrial IoT. IEEE Transactions on Industrial Informatics.
Zhao, H., Sun, S., & Jin, B. (2018). Sequential Fault Diagnosis Based on LSTM Neural Network. IEEE ACCESS, 6, 12929-12939.









中文文獻
[1] 張耀仁，「太陽能板的製程對於環境的衝擊」，工研院核能研究所， 2017。
[2] 洪春暉，「CPS於智慧製造之應用探索」，工研院產研所， 2016。
[3] 黃偉碩，「利用貝氏分類與因數分析法於半導體製程錯誤偵測與診斷」， 中華大學科技管理學系碩士論文，2005。
[4] 張庭嘉，「建構半導體黃光製程異常升溫現象之模糊監控技術」，國立清華大學工業工程與工程管理學系碩士論文，2007。
[5] 劉淑範，「半導體黃光製程機台異常升溫曲線偵測之研究」，行政院國家科學委員會補助專題研究計畫，2008。
[6] 曾介亭，「多代理人系統在建構APC作業平台之研究」，國立成功大學工程科學系專班碩士論文，2006。
[7] 戴宏恩，「核糖核酸與蛋白質結合位置之特徵探勘與預測」，長庚大學資訊工程學系碩士論文，2015。
[8] 林嘉瑩，「應用於半導體製程資料分析之正規化機器學習方法」，國立交通大學電子工程學系碩士論文，2016。
[9]戴鴻恩，「利用半導體製程即時參數值建立缺陷特徵化監控模型」，國立台灣科技大學管理研究所碩士論文，2005。