為協助操作員監控核電廠(Nuclear Power Plant, NPP)的運轉狀態以確保運轉安全，在電廠發生異常事件時，能即時偵測並辨別出該事件之類型，是不可忽視的議題。若該異常事件是從未見過的未知事件，更需進一步隔離出，避免操作員因誤判異常事件，從而導致錯誤操作。本研究因應此議題提出一套未知事件隔離系統，能夠自動偵測事件發生與辨識事件類型，用以輔助核電操作員做出決策。建立正常狀態下的LSTM模型(Long Short-Term Memory, LSTM)得到相對應的預測值，通過與事件本身真實值相對比設定正常波動範圍的域值以此監測電廠的運行狀態。在偵測到事件發生後，通過LSTM分類模型自主學習各類事件的特徵達到區別開各類事件類別的目的。再者由於事件類別繁多，在現有研究中大多是針對高發性事件的辨識系統，並未包括實際發生概率較少的類別，故而本研究再提出未知事件隔離系統，以避免誤判。本論文所發展的方法，將以馬鞍山核電廠模擬程式PCTran（Personal Computer Transient Analyzer，PCTran）-PWR（Pressurized water reactor，PWR）所產生模擬數據來進行效能驗證，實驗結果證明瞭所提方案的有效性。

To assist the operator in monitoring the status of the nuclear power plant (NPP) to ensure safe operation, detecting and identifying an abnormal event cannot be ignored. If the abnormal event is an event that has never been seen (i.e., an unknown event), it needs to be further isolated to prevent the operator from taking improper recovery actions due to misjudging the abnormal event. In response to this issue, this study proposes an unknown event isolation system that can automatically detect the occurrence of an event and identify its type, and isolate an abnormal event. The predicted value obtained by the LSTM model and the true value of the event are used to set the threshold for the normal state. After detecting an event, the characteristics of events are learned independently through the LSTM classification model to achieve the purpose of distinguishing various types of events. In addition, due to a large number of event categories, most of the existing studies are identification systems for frequent events, which do not include categories that have a low probability of occurring. Therefore, this study proposes an unknown event isolation system to avoid this misjudgment. The methods developed in this paper will be evaluated using the data generated by the simulation program PCTran-PWR of the Maanshan NPP, and the experimental results demonstrate the efficacy of the proposed schemes.

摘要 I
ABSTRACT II
誌謝 III
目錄 IV
圖目錄 VI
表目錄 VIII
名詞縮寫表 IX
第一章 緒論 1
1.1 研究動機 1
1.2 研究方法與文獻回顧 2
1.3 研究架構 5
第二章 研究方法 7
2.1 深度學習 7
2.2 長短期記憶 9
2.3 自動編碼器 12
第三章 系統設計 14
3.1 事件偵測 14
3.2 事件辨識 16
3.3 事件隔離 17
第四章 數據模擬與設定 19
4.1 數據模擬 19
4.2 數據形式與設定 24
第五章 系統結果 26
5.1 偵測結果 26
5.2 辨識結果 27
5.3 隔離結果 29
第六章 未知事件辨別 36
6.1 系統驗證 36
6.2 結果與討論 37
第七章 總結 44
參考文獻 45

[1] Y. Bartal, J. Lin, and R. E. Uhrig, "Nuclear power plant transient diagnostics using artificial neural networks that allow “don’t-know” classifications," Nucl. Technol. 110, 436-449, 1995.
[2] K. Hadad, M. Mortazavi, M. Mastali, and A. A. Safavi, "Enhanced neural network based fault detection of a VVER nuclear power plant with the aid of principal component analysis," IEEE Trans. Nucl. Sci. 55, 3611-3619, 2008.
[3] F.-H. Hwang and S.-L. Hwang, "Design and evaluation of computerized operating procedures in nuclear power plants," Ergonomics 46, 271-284, 2003.
[4] J. M. Broughton, P. Kuan, D. A. Petti, and E. Tolman, "A scenario of the Three Mile Island unit 2 accident," Nucl. Technol. 87, 34-53, 1989.
[5] D.-Y. Kim, "Cyber security issues imposed on nuclear power plants," Ann. Nucl. Energy. 65, 141-143, 2014.
[6] A. Keliris, H. Salehghaffari, B. Cairl, P. Krishnamurthy, M. Maniatakos, and F. Khorrami, "Machine learning-based defense against process-aware attacks on industrial control systems," in 2016 IEEE International Test Conference (ITC), 2016: IEEE,1-10.
[7] J. P. Farwell and R. Rohozinski, "Stuxnet and the future of cyber war," Survival 53, 23-40, 2011.
[8] H. S. Cho and T. H. Woo, "Cyber security in nuclear industry–Analytic study from the terror incident in nuclear power plants (NPPs)," Ann. Nucl. Energy. 99, 47-53, 2017.
[9] P. F. Ikonomou, Global Nuclear Developments: Insights from a Former IAEA Nuclear Inspector, Springer Nature.
[10] C. Yin, S. Zhang, J. Wang, and N. N. Xiong, "Anomaly Detection Based on Convolutional Recurrent Autoencoder for IoT Time Series," IEEE Trans. Syst. Man Cy-S, 2020.
[11] F. Giannoni, M. Mancini, and F. Marinelli, "Anomaly detection models for IoT time series data," arXiv preprint arXiv:1812.00890, 2018.
[12] T.-H. Lin, S.-C. Wu, and H.-P. Chou, "A Novel Feature Extraction Scheme for NPP Initiating Event Identification," in International Conference on Nuclear Engineering, 2016, vol. 50015: American Society of Mechanical Engineers, V001T04A013.
[13] P. Smyth, "Hidden Markov models for fault detection in dynamic systems," Patternrecogn. 27, 149-164, 1994.
[14] L. Huijuan, C. Jianguo, and W. Wei, "Two stratum bayesian network based anomaly detection model for intrusion detection system," in 2008 International Symposium on Electronic Commerce and Security, 2008: IEEE, 482-487.
[15] J. Rrushi and R. Campbell, "Detecting cyber attacks on nuclear power plants," in International Conference on Critical Infrastructure Protection, 2008: Springer, 41-54.
[16] A. Almalawi, X. Yu, Z. Tari, A. Fahad, and I. Khalil, "An unsupervised anomaly-based detection approach for integrity attacks on SCADA systems," Comput. Secur. 46, 94-110, 2014.
[17] J. Liu, R. Seraoui, V. Vitelli, and E. Zio, "Nuclear power plant components condition monitoring by probabilistic support vector machine," Ann. Nucl. Energy. 56, 23-33, 2013.
[18] K. Hadad, M. Pourahmadi, and H. Majidi-Maraghi, "Fault diagnosis and classification based on wavelet transform and neural network," Prog. Nucl. Energ. 53, 41-47, 2011.
[19] J. Ma and J. Jiang, "Applications of fault detection and diagnosis methods in nuclear power plants: A review," Prog. Nucl. Energ. 53, 255-266, 2011.
[20] J. J. Gertler, "Survey of model-based failure detection and isolation in complex plants," IEEE Contr. Syst. Mag. 8, 3-11, 1988.
[21] R. Isermann, "Fault diagnosis of machines via parameter estimation and knowledge processing—tutorial paper," Automatica. 29, 815-835, 1993.
[22] J. Calado, J. Korbicz, K. Patan, R. J. Patton, and J. S. Da Costa, "Soft computing approaches to fault diagnosis for dynamic systems," Eur. J. Control. 7, 248-286, 2001.
[23] V. Palade and C. D. Bocaniala, Computational intelligence in fault diagnosis, Springer, 2006.
[24] J. Korbicz, J. M. Koscielny, Z. Kowalczuk, and W. Cholewa, Fault diagnosis: models, artificial intelligence, applications, Springer Science & Business Media, 2012.
[25] R. Razavi-Far, H. Davilu, V. Palade, and C. Lucas, "Model-based fault detection and isolation of a steam generator using neuro-fuzzy networks," Neurocomputing 72, 2939-2951, 2009.
[26] M. G. Na et al., "Prediction of major transient scenarios for severe accidents of nuclear power plants," IEEE Trans. Nucl. Sci. 51, 313-321, 2004.
[27] M. J. Embrechts and S. Benedek, "Hybrid identification of nuclear power plant transients with artificial neural networks," IEEE Trans. Ind. Electron. 51, 686-693, 2004.
[28] F. Jamil, M. Abid, M. Adil, I. Haq, A. Khan, and S. Khan, "Kernel approaches for fault detection and classification in PARR-2," J. Process Contr. 64, 1-6, 2018.
[29] K.-C. Kwon, "HMM-based transient identification in dynamic process," Int. J. Control, Autom. Syst. 2, 40-46, 2000.
[30] D. Roverso, "Plant diagnostics by transient classification: The aladdin approach," Int. J. Int. Syst. 17, 767-790, 2002.
[31] M. G. Na, W. S. Park, and D. H. Lim, "Detection and diagnostics of loss of coolant accidents using support vector machines," IEEE Trans. Nucl. Sci. 55, 628-636, 2008.
[32] B.-S. Peng, H. Xia, Y.-K. Liu, B. Yang, D. Guo, and S.-M. Zhu, "Research on intelligent fault diagnosis method for nuclear power plant based on correlation analysis and deep belief network," Prog. Nucl. Energ. 108, 419-427, 2018.
[33] F.-C. Chen and M. R. Jahanshahi, "NB-CNN: Deep learning-based crack detection using convolutional neural network and Naïve Bayes data fusion," IEEE Trans. Ind. Electron. 65, 4392-4400, 2017.
[34] S. Şeker, E. Ayaz, and E. Türkcan, "Elman's recurrent neural network applications to condition monitoring in nuclear power plant and rotating machinery," Eng. Appl. Artif. Intel. 16, 647-656, 2003.
[35] Y. LeCun, Y. Bengio, and G. Hinton, "Deep learning," Nature. 521, 436-444, 2015.
[36] G. Dong and H. Liu, Feature engineering for machine learning and data analytics, CRC Press, 2018.
[37] 斎藤康毅, ゼロから作る Deep Learning: Python で学ぶディープラーニングの理論と実装. オライリー・ジャパン, 2016.
[38] S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Comput. 9, 1735-1780, 1997.
[39] Y.-L. Kong, Q. Huang, C. Wang, J. Chen, J. Chen, and D. He, "Long short-term memory neural networks for online disturbance detection in satellite image time series," Remote Sens. 10, 452, 2018.
[40] S. Naseer et al., "Enhanced network anomaly detection based on deep neural networks," IEEE Access. 6, 48231-48246, 2018.
[41] E. Marchi, F. Vesperini, F. Eyben, S. Squartini, and B. Schuller, "A novel approach for automatic acoustic novelty detection using a denoising autoencoder with bidirectional LSTM neural networks," in Proceedings 40th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2015, 2015, 5.
[42] J. Yang, S. Lee, and J. Kim, "Nuclear Power Plant Accident Diagnosis Algorithm Including Novelty Detection Function Using LSTM," in International Conference on Applied Human Factors and Ergonomics, 2019: Springer, 644-655.
[43] J. Inoue, Y. Yamagata, Y. Chen, C. M. Poskitt, and J. Sun, "Anomaly detection for a water treatment system using unsupervised machine learning," in 2017 IEEE International Conference on Data Mining Workshops (ICDMW), 2017: IEEE, 1058-1065.