為確保核電廠於運轉期間安全無虞，在遭遇異常情況時，我們必須能迅速診斷出異常的肇始事件，並進一步評估該事件(如：冷卻水流失事件)的嚴重程度。為達到此目的，本論文嘗試應用深度學習技術來建立一套電廠運轉監控系統，用以處理電廠面對肇始事件時，所衍伸出的三大議題：「肇始事件辨識」、「肇始事件嚴重程度之評估」與「肇始事件情狀下進行失能感測器的訊號重建」。
當肇始事件發生時，此系統採用過去我們實驗室的兩篇碩士論文工作，使用統計方法Hotelling’s 統計來進行事件偵測的工作。事件辨識的部分，首在擷取偵測到異常後60秒的感測器量測讀值，送入卷積神經網路進行辨識工作。若辨識出的事件為冷卻水流失事件、主蒸汽管斷裂事件及蒸汽產生器破管事件，則須進一步評估破口面積。此外，若事件發生前已知有失能感測器的存在，我們結合多變量自回歸模型的想法，融入至卷積神經網路，使神經網路有能力利用過去時間點的數據，來預測失能感測器當下的讀值。
本論文所發展的方法，將以核三廠電廠模擬程式PCTran-PWR所產生肇始事件數據來進行效能驗證。透過卷積神經網路對測試數據進行辨識，可有近100 %的高辨識率；事件嚴重程度評估的部分，平均絕對百分比誤差都在2%左右；考慮各感測器在各肇始事件的情況下進行重建工作，在494個案例中計有408個案例可達相對誤差低於0.1以下，比率超過82%。

In order to ensure the safety of the nuclear power plant (NPP), it is necessary to identify the initial event (IE) when abnormal situation occurs and then predict the severity of the IE (e.g. loss of coolant accident, LOCA). In this study, we attempt to apply the deep learning technique to build up an NPP operation monitoring system, which is used to solve three issues when dealing with an IE: IE identification, IE severity prediction, and signal reconstruction (SR) for the faulty sensor.
To begin, the system adopts the Hotelling’s test to detect a unknown IE. As for IE identification, we capture sensor signals for 60 seconds after event detection as the input for the convolution neural network (CNN) to identify the IE. If the result of identification is LOCA, main steam line break accident or steam generator tube rupture, the system will further predict thesize of break area. Finally, when there is a known faulty sensor, we integrate ideas of multivariable autoregressive model into CNN such that the resulting network model is able to predict future sensor readings by using its past readings .
The proposed method is verified by the IE data generated by the Maanshan NPP simulator. From the results, an identification rate close to 100 % is obtained by the proposed CNN model. The mean absolute percentage errors of the severity prediction are all about 2 %. Among the 494 SR cases, there are more than 82% of them having a relative error less than 0.1.

1.CNSC, "CNSC Fukushima Task Force Report," Catalogue number INFO-0824, October 2011 2011.
2.IAEA, "Defining initiating events for purposes of probabilistic safety assessment," IAEA-TECDOC-719, 1993.
3.K. Henning, "Recommendations for implementing the strategic initiative INDUSTRIE 4.0," 2013.
4.K.-C. Kwon and J.-H. Kim, "Accident identification in nuclear power plants using hidden Markov models," Engineering Applications of Artificial Intelligence, vol. 12, no. 4, pp. 491-501, 1999/08/01/ 1999.
5.N. Man Gyun et al., "Prediction of major transient scenarios for severe accidents of nuclear power plants," IEEE Transactions on Nuclear Science, vol. 51, no. 2, pp. 313-321, 2004.
6.Y.-G. K. No, Ju-Hyun ; Na, Man-Gyun ; Lim, Dong-Hyuk ; Ahn, Kwang-Il, "MONITORING SEVERE ACCIDENTS USING AI TECHNIQUES," Nuclear Engineering and Technology, vol. 44, no. 4, pp. 393-404, 2012.
7.K. Kwon, "HMM-based transient identification in dynamic process," Transactions on Control, Automation, and Systems Engineering, vol. 2, pp. 40-46, 2000.
8.J. A. Carlos Canedo Medeiros and R. Schirru, "Identification of nuclear power plant transients using the Particle Swarm Optimization algorithm," Annals of Nuclear Energy, vol. 35, no. 4, pp. 576-582, 2008/04/01/ 2008.
9.A. d. S. Nicolau, R. Schirru, and A. Alvarenga de Moura Meneses, "Quantum evolutionary algorithm applied to transient identification of a nuclear power plant," Progress in Nuclear Energy, vol. 53, no. 1, pp. 86-91, 2011/01/01/ 2011.
10.M. Marseguerra, E. Zio, A. Oldrini, and E. Brega, "Fuzzy identification of transients in nuclear power plants," Nuclear Engineering and Design, vol. 225, no. 2, pp. 285-294, 2003/11/01/ 2003.
11.D. Roverso, "Soft computing tools for transient classification," Information Sciences, vol. 127, no. 3, pp. 137-156, 2000/08/01/ 2000.
12.T.-H. Lin, S.-C. Wu, K.-Y. Chen, and H.-P. Chou, "Feature extraction and sensor selection for NPP initiating event identification," Annals of Nuclear Energy, vol. 103, pp. 384-392, 2017/05/01/ 2017.
13.NRC, "Anticipated transients without scram for light water reactors," NUREG-0460, 1978.
14.陳光佑, "壓水式反應器肇始事件偵測與辨識之研究," 2017.
15.K. Moshkbar-Bakhshayesh and M. B. Ghofrani, "Transient identification in nuclear power plants: A review," Progress in Nuclear Energy, vol. 67, pp. 23-32, 2013/08/01/ 2013.
16.D. Roverso, "Plant Diagnostics by TransientClassification: The ALADDIN Approach," International Journal of Intelligent Systems, vol. 17, no. 8, pp. 767-790, 2002.
17.M. J. Embrechts and S. Benedek, "Hybrid identification of nuclear power plant transients with artificial neural networks," IEEE Transactions on Industrial Electronics, vol. 51, no. 3, pp. 686-693, 2004.
18.T. V. Santosh, G. Vinod, R. K. Saraf, A. K. Ghosh, and H. S. Kushwaha, "Application of artificial neural networks to nuclear power plant transient diagnosis," Reliability Engineering & System Safety, vol. 92, no. 10, pp. 1468-1472, 2007/10/01/ 2007.
19.K. Mo, S. J. Lee, and P. H. Seong, "A dynamic neural network aggregation model for transient diagnosis in nuclear power plants," Progress in Nuclear Energy, vol. 49, no. 3, pp. 262-272, 2007/04/01/ 2007.
20.M. G. Na, W. S. Park, and D. H. Lim, "Detection and Diagnostics of Loss of Coolant Accidents Using Support Vector Machines," IEEE Transactions on Nuclear Science, vol. 55, no. 1, pp. 628-636, 2008.
21.S. Benedek and M. Embrechts, "Rapid identification of nuclear power plant malfunctions with artificial neural networks via Fourier transformed signals," in Proceedings of ANNNIE'96, St. Louis Missouri, 1996, pp. 903-909.
22.I. Jolliffe, "Principal component analysis," Wiley Online Library, 2002.
23.X. Tian, V. Becerra, N. Bausch, T. V. Santhosh, and G. Vinod, "A study on the robustness of neural network models for predicting the break size in LOCA," Progress in Nuclear Energy, vol. 109, pp. 12-28, 2018/11/01/ 2018.
24.M. Saghafi and M. B. Ghofrani, "Real-time estimation of break sizes during LOCA in nuclear power plants using NARX neural network," Nuclear Engineering and Technology, vol. 51, no. 3, pp. 702-708, 2019/06/01/ 2019.
25.M. A. Kramer, "Autoassociative neural networks," Computers & Chemical Engineering, vol. 16, no. 4, pp. 313-328, 1992/04/01/ 1992.
26.R. Dunia, Qin, S.J., Edgar, T.F., McAvoy, T.J., "Identification of faulty sensors using principal component analysis," Process Systems Engineering, vol. 42, no. 10, 1996.
27.P. Baraldi, A. Cammi, F. Mangili, and E. Zio, "An ensemble approach to sensor fault detection and signal reconstruction for nuclear system control," Annals of Nuclear Energy, vol. 37, no. 6, pp. 778-790, 2010/06/01/ 2010.
28.W. Li, M. Peng, Y. Liu, N. Jiang, H. Wang, and Z. Duan, "Fault detection, identification and reconstruction of sensors in nuclear power plant with optimized PCA method," Annals of Nuclear Energy, vol. 113, pp. 105-117, 2018/03/01/ 2018.
29.K. R. Uren, G. van Schoor, C. P. du Rand, and A. Botha, "An integrated approach to sensor FDI and signal reconstruction in HTGRs – Part I: Theoretical framework," Annals of Nuclear Energy, vol. 87, pp. 750-760, 2016/01/01/ 2016.
30.K. R. Uren, G. van Schoor, C. P. du Rand, and A. Botha, "An integrated approach to sensor FDI and signal reconstruction in HTGRs – Part II: Case studies," Annals of Nuclear Energy, vol. 87, pp. 739-749, 2016/01/01/ 2016.
31.T.-H. Lin and S.-C. Wu, "Sensor fault detection, isolation and reconstruction in nuclear power plants," Annals of Nuclear Energy, vol. 126, pp. 398-409, 2019/04/01/ 2019.
32.斎藤康毅, "Deep Learning：用Python進行深度學習的基礎理論實作," 2017.
33.S. J. a. p. a. Ruder, "An overview of gradient descent optimization algorithms," 2016.
34.Y. J. I. T. o. A. C. Chien, "Pattern classification and scene analysis," vol. 19, no. 4, pp. 462-463, 1974.
35.Y. LeCun, Y. Bengio, and G. J. n. Hinton, "Deep learning," vol. 521, no. 7553, p. 436, 2015.
36.A. F. J. a. p. a. Agarap, "Deep learning using rectified linear units (relu)," 2018.
37.A. Krizhevsky, I. Sutskever, and G. E. Hinton, "Imagenet classification with deep convolutional neural networks," in Advances in neural information processing systems, 2012, pp. 1097-1105.
38.L. lab. ("Convolution arithmetic tutorial -- Theano 1.0.0 documentation. " http://deeplearning.net/software/theano/tutorial/conv_arithmetic.html (accessed Dec. 18, 2019)).
39.A. Gulli and S. Pal, Deep Learning with Keras. Packt Publishing Ltd, 2017.
40.N. E. West and T. O'Shea, "Deep architectures for modulation recognition," in 2017 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), 2017, pp. 1-6: IEEE.