AI概念提出至今已超過一甲子，但受限於晶片運算及記憶體等限制以致人們寄望AI能夠達成的許多應用無法實現，然而近年來隨著硬體技術發展一日千里，深度學習演算法逐漸受到工業界重視，透過即時監控及巨量資料分析朝向智慧製造與工業4.0的方向邁進；現今工業幾乎已全面自動化生產，各個設備中皆安裝不同種類的傳感器以利操作員追蹤工廠生產狀態，在這些看似無意義的歷史數據當中卻有可能隱藏著能夠優化工廠的重要信息，透過對海量歷史數據進行分析並建立機器學習模型將會比建立傳統的知識模型(knowledge-based model)更有效率及準確率。
本研究收集了燃氣複合循環電廠(Combined Cycle Power Plant)之歷年環境條件、滿載發電量、各設備溫度壓力和燃氣渦輪排放等資料建立滿載發電量及汙染排放之迴歸模型，以及使用田納西伊士曼程序(Tennessee Eastman Process)之故障模擬資料建立應用於化工程序上之故障檢測模型，透過引入AI的方式提供優化實際工廠之經濟面、環境面及安全面的解決方案。為了建立高精確度模型而使用python進行機器學習，首先將收集到的數據標準化(Standardization)使各項變數之數值壓縮在相似區間並劃分成訓練集、驗證集及測試集三個部分，再使用斯皮爾曼相關係數分析來選擇輸入變數，接著分別訓練類神經網路、長短期記憶、隨機森林及極限梯度提升四種機器學習模型並透過驗證集資料進行超參數調整提升模型準確率，最後透過測試集資料當作未知數據測試模型以增加模型在面對未來資料的可信度，藉由相關文獻的比較本研究所提出的數據處理流程及機器學習模型表現相對優秀，未來將有望實際應用於工業界當中。

The concept of AI was proposed a long time ago, but many applications cannot achieve due to the limitations of chip computing and memory. However, in recent years, with the rapid development of hardware technology, deep learning have gradually attract the attention of the industry. Through real-time monitoring and big data analysis, move towards smart manufacturing and Industry 4.0. Nowadays, the industry has almost fully automated production. Different types of sensors are installed in various equipment to track the production of the factory. These seemingly meaningless historical data may hide important information that can optimize the factory. Analyzing massive historical data and building machine learning models will be more efficient and accurate than building traditional knowledge-based model.
This study collected data on environmental conditions, full-load power generation, temperature and pressure of various equipment, and gas turbine emissions of gas-fired combined cycle power plants to establish a regression model for full-load power generation and pollution emissions, and use the fault simulation data of the Tennessee Eastman Process to establish a fault detection model applied to the chemical process. Provide solutions that optimize the economic, environmental and safety aspects of the actual factory by introducing AI.
In order to build a high-precision model and use python for machine learning. First, standardize the data to compressed the value to the similar interval and divided the data into three parts: training set, validation set and test set. Then use Spearman’s correlation coefficient analysis to select the input variables, and then train the four machine learning models of Neural Network, Long Short-Term Memory, Random Forest, and XGBoost respectively, and use the validation data to adjust the hyperparameters to improve the accuracy of the model. Finally, the test set data is used as an unknown data test model to increase the model’s credibility in the future data. Based on the comparison of relevant literature, the data processing flow and machine learning model proposed by this research are relatively excellent, and it is expected to be applied in the industry in the future.

[1] A.A.Islikaye, A.Cetin, Performance of ML methods in estimating net energy produced in a combined cycle power plant, Proc. - 2018 6th Int. Istanbul Smart Grids Cities Congr. Fair, ICSG 2018. (2018) 217–220.
[2] 李沛濠, 廖孟儀, 馬鴻文, 李孟穎, 洪明龍, 臺灣火力發電健康衝擊外部成本分析, J. Taiwan Energy. 3 (2016) 277–292.
[3] V.Venkatasubramanian, The promise of artificial intelligence in chemical engineering: Is it here, finally?, AIChE J. 65 (2019) 466–478.
[4] L.H.Chiang, E.Russell, R.D.Braatz, Fault Detection and Diagnosis in Industrial Systems., 2001.
[5] J.J.Downs, E.F.Vogel, A plant-wide industrial process control problem, Comput. Chem. Eng. 17 (1993) 245–255.
[6] P.Tüfekci, Prediction of full load electrical power output of a base load operated combined cycle power plant using machine learning methods, Int. J. Electr. Power Energy Syst. 60 (2014) 126–140.
[7] A.Rezazadeh, Environmental pollution prediction of NOx by predictive modelling and process analysis in natural gas turbine power plants, Pollution. 7 (2021) 481–494.
[8] S.Elfwing, E.Uchibe, K.Doya, Sigmoid-weighted linear units for neural network function approximation in reinforcement learning, Neural Networks. 107 (2018) 3–11.
[9] D.P.Kingma, J.L.Ba, Adam: A method for stochastic optimization, 3rd Int. Conf. Learn. Represent. ICLR 2015 - Conf. Track Proc. (2015) 1–15.
[10] S.Hochreiter, J.Schmidhuber, Long Short-Term Memory, Neural Comput. 9 (1997) 1735–1780.
[11] P.Perner, LNAI 7376 - Machine Learning and Data Mining in Pattern Recognition,2012.
[12] R.J.Lewis, D.Ph, W.C.Street, An Introduction to Classification and Regression Tree ( CART ) Analysis, (2000).
[13] P.Fuleky, Macroeconomic Forecasting in the Era of Big Data, 2020.
[14] J.Jiang, B.Cui, C.Zhang, F.Fu, DimBoost: Boosting gradient boosting decision tree to higher dimensions, Proc. ACM SIGMOD Int. Conf. Manag. Data. (2018) 1363–1376.
[15] A.Agarwal, P.Barham, E.Brevdo, Z.Chen, C.Citro, G.S.Corrado, A.Davis, J.Dean, M.Devin, S.Ghemawat, I.Goodfellow, A.Harp, G.Irving, M.Isard, Y.Jia, R.Jozefowicz, L.Kaiser, M.Kudlur, J.Levenberg, D.Man, R.Monga, S.Moore, D.Murray, C.Olah, M.Schuster, J.Shlens, B.Steiner, I.Sutskever, K.Talwar, P.Tucker, V.Vanhoucke, V.Vasudevan, F.Vi, O.Vinyals, P.Warden, M.Wattenberg, M.Wicke, Y.Yu, X.Zheng, TensorFlow: Large-Scale Machine Learning on Heterogeneous Distributed Systems, (2015).
[16] Y.Wang, Z.Pan, X.Yuan, C.Yang, W.Gui, A novel deep learning based fault diagnosis approach for chemical process with extended deep belief network, ISA Trans. 96 (2020) 457–467.
[17] C.A.Rieth, B.D.Amsel, R.Tran, M.B.Cook, Issues and advances in anomaly detection evaluation for joint human-automated systems, Adv. Intell. Syst. Comput. 595 (2018) 52–63.
[18] Y.Zhang, Enhanced statistical analysis of nonlinear processes using KPCA , KICA and SVM, 64 (2009).
[19] M.Shams, H.Budman, T.Duever, Fault detection using CUSUM based techniques with application to the Tennessee Eastman Process, IFAC, 2010.
[20] M.Grbovic, W.Li, N.A.Subrahmanya, A.K.Usadi, S.Vucetic, Cold start approach for data-driven fault detection, IEEE Trans. Ind. Informatics. 9 (2013) 2264–2273.
[21] D.Xie, L.Zhang, L.Bai, Deep Learning in Visual Computing and Signal Processing, Appl. Comput. Intell. Soft Comput. 2017 (2017).
[22] L.H.Chiang, M.E.Kotanchek, A.K.Kordon, Fault diagnosis based on Fisher discriminant analysis and support vector machines, Comput. Chem. Eng. 28 (2004) 1389–1401.
[23] A.Kulkarni, V.K.Jayaraman, B.D.Kulkarni, Knowledge incorporated support vector machines to detect faults in Tennessee Eastman Process, Comput. Chem. Eng. 29 (2005) 2128–2133.
[24] R.Eslamloueyan, Designing a hierarchical neural network based on fuzzy clustering for fault diagnosis of the Tennessee-Eastman process, Appl. Soft Comput. J. 11 (2011) 1407–1415.
[25] S.Yin, S.X.Ding, A.Haghani, H.Hao, P.Zhang, A comparison study of basic data-driven fault diagnosis and process monitoring methods on the benchmark Tennessee Eastman process, J. Process Control. 22 (2012) 1567–1581.
[26] C.Jing, J.Hou, SVM and PCA based fault classification approaches for complicated industrial process, Neurocomputing. 167 (2015) 636–642.
[27] D.Xie, L.Bai, A hierarchical deep neural network for fault diagnosis on Tennessee-Eastman process, Proc. - 2015 IEEE 14th Int. Conf. Mach. Learn. Appl. ICMLA 2015. (2016) 745–748.
[28] D.Xie, L.Bai, A hierarchical deep neural network for fault diagnosis on Tennessee-Eastman process, Proc. - 2015 IEEE 14th Int. Conf. Mach. Learn. Appl. ICMLA 2015. (2016) 745–748.
[29] Z.Zhang, J.Zhao, A deep belief network based fault diagnosis model for complex chemical processes, Comput. Chem. Eng. 107 (2017) 395–407.
[30] S.Heo, J.H.Lee, Fault detection and classification using artificial neural networks, IFAC-PapersOnLine. 51 (2018) 470–475.
[31] S.Alketbi, A.B.Nassif, M.A.Eddin, I.Shahin, A.Elnagar, Predicting the power of a combined cycle power plant using machine learning methods, Proc. 2020 IEEE Int. Conf. Commun. Comput. Cybersecurity, Informatics, CCCI 2020. (2020).
[32] E.U.Heysem KAYA,Pinar TUFEKCI, Predict CO and NOx emission from gas turbine.pdf, Turkish J. Electr. Eng. Comput. Sci. (2019) 4738–4796.
[33] A.Ajagekar, F.You, Quantum computing assisted deep learning for fault detection and diagnosis in industrial process systems, Comput. Chem. Eng. 143 (2020).
[34] Y.Wang, D.Wu, X.Yuan, LDA-based deep transfer learning for fault diagnosis in industrial chemical processes, Comput. Chem. Eng. 140 (2020).
[35] F.Lv, C.Wen, Z.Bao, M.Liu, Fault diagnosis based on deep learning, Proc. Am. Control Conf. 2016-July (2016) 6851–6856.
[36] D.Xie, L.Zhang, L.Bai, Deep Learning in Visual Computing and Signal Processing, Appl. Comput. Intell. Soft Comput. 2017 (2017).
[37] A.Kulkarni, V.K.Jayaraman, B.D.Kulkarni, Knowledge incorporated support vector machines to detect faults in Tennessee Eastman Process, Comput. Chem. Eng. 29 (2005) 2128–2133.
[38] M.Grbovic, W.Li, N.A.Subrahmanya, A.K.Usadi, S.Vucetic, Cold start approach for data-driven fault detection, IEEE Trans. Ind. Informatics. 9 (2013) 2264–2273.
[39] S.Chen, J.Yu, S.Wang, One-dimensional convolutional auto-encoder-based feature learning for fault diagnosis of multivariate processes, J. Process Control. 87 (2020) 54–67.
[40] C.Tong, X.Yan, Statistical process monitoring based on a multi-manifold projection algorithm, Chemom. Intell. Lab. Syst. 130 (2013) 20–28.
[41] P.Wu, L.Guo, S.Lou, J.Gao, Local and Global Randomized Principal Component Analysis for Nonlinear Process Monitoring, IEEE Access. 7 (2019) 25547–25562.
[42] E.A.Elfaki, A.H.Ahmed, Prediction of Electrical Output Power of Combined Cycle Power Plant Using Regression ANN Model, J. Power Energy Eng. 06 (2018) 17–38.
[43] P.Tü, S.Gürgen, Local and Global Learning Methods for Predicting Power of a Combined Gas & Steam Turbine, (2014).
[44] M.Si, T.J.Tarnoczi, B.M.Wiens, K.Du, Development of Predictive Emissions Monitoring System Using Open Source Machine Learning Library-Keras: A Case Study on a Cogeneration Unit, IEEE Access. 7 (2019) 113463–113475.
[45] J.B.O.Mitchell B.O., Machine learning methods in chemoinformatics, Wiley Interdiscip. Rev. Comput. Mol. Sci. 4 (2014) 468–481.
[46] D.Michie, Methodologies from machine learning in data analysis and software, Comput. J. 34 (1991) 559–565.
[47] M.Lewis-Beck, A.Bryman, T.Futing Liao, Canonical Correlation Analysis, SAGE Encycl. Soc. Sci. Res. Methods. (2012) 3–5.
[48] Y.Lei, B.Yang, X.Jiang, F.Jia, N.Li, A.K.Nandi, Applications of machine learning to machine fault diagnosis: A review and roadmap, Mech. Syst. Signal Process. 138 (2020) 106587.
[49] I.Martin-Diaz, D.Morinigo-Sotelo, O.Duque-Perez, R.J.Romero-Troncoso, An Experimental Comparative Evaluation of Machine Learning Techniques for Motor Fault Diagnosis under Various Operating Conditions, IEEE Trans. Ind. Appl. 54 (2018) 2215–2224.
[50] Z.Chen, K.Gryllias, W.Li, Mechanical fault diagnosis using Convolutional Neural Networks and Extreme Learning Machine, Mech. Syst. Signal Process. 133 (2019) 106272.
[51] V.Venkatasubramanian, K.Chan, A neural network methodology for process fault diagnosis, AIChE J. 35 (1989) 1993–2002.
[52] P.K.Wong, Z.Yang, C.M.Vong, J.Zhong, Real-time fault diagnosis for gas turbine generator systems using extreme learning machine, Neurocomputing. 128 (2014) 249–257.
[53] P.Langley, H.A.Simon, Applications of Machine Learning and Rule Induction, Commun. ACM. 38 (1995) 54–64.
[54] Z.Ge, Z.Song, S.X.Ding, B.Huang, 过程工业中的数据挖掘和分析:机器学习的作用, IEEE Access. 5 (2017) 20590–20616.
[55] I.Arel, D.C.Rose, T.P.Karnowski, Deep Machine Learning — A New Frontier in Artificial Intelligence Research, (2010) 13–18.
[56] Jay Lee (李傑), Hossein Davari, Jaskaran Singh, Vibhor Pandhare. 李杰：工业人工智能与工业4.0 制造. SME Manufacturing Letter 18 （2018）20-23.