為協助核電廠運轉員及早辨認出肇始事件，以做出對應的措施來降低嚴重事故發生的可能性，本研究提出一套事故偵測與辨識演算法，能夠自動偵測事故的發生並準確的辨識出事故的類型，使有效維持核電廠安全運轉。藉由監測核電廠的感測訊號，當其讀值超出預先設定之界限時，即可偵測出事故的發生。與以前做法不同的是，本研究採用統計方法來定出客觀合理的界限。事故辨識的問題相當於機器學習中圖型辨識問題。在偵測到事故發生後，辨識系統會先擷取當下訊號中重要的特徵，用以與資料庫中所有事故類別進行比對，以找出與該事故最為相似的類別，以作為該未知事故的種類。然而在過去的事故辨識方法裡，其所擷取的特徵皆以時域資訊為主，而空間資訊則是被忽略。故而，本研究將著重在空間資訊於事件辨識的應用，並與時域資訊的成效做比較。再者，為了提高辨識成功率，對擷取的特徵依照其事件區分能力進行排序，以此挑選出對辨識有幫助的特徵值，使後續分類更加容易。最後，在辨識時可能會遇到事故屬於沒有建檔的類型，為因應此狀況，本研究提出一個能在辨識類別前有效將此種事故先隔離出來的方法，以避免由於未知類別所導致的誤判。在本研究中所有提出的演算法，將以核三廠壓水式反應器¬的模擬器Modular Accident Analysis Program Version 5 (MAAP5)所生成的肇始事件資料來進行驗證，總計11類135個肇始事件將用來試驗我們所提出之事故辨識系統的效率與穩健性。

To help operators in nuclear power plant (NPP) identifying an initiating event and take proper actions to avoid the occurrence of a severe accident, several event detection and identification algorithms are proposed in this study. These algorithms can automatically detect the occurrence of an event and identify its type to enable a safe shutdown of the NPP. By monitoring various sensing variables, an event can be detected when the readings exceed the preset limits. Unlike methods in the literature, a statistical approach is applied to set the desired thresholds objectively. After detecting an event, features that are discriminant will be extracted to compare against those of all the events stored in the database to determine its type. In the existing approaches, the features are extracted with emphasis on their temporal aspects, ignoring the spatial information. Thus, this study would focus on the application of the spatial information in event identification. Moreover, to increase the success rate of event identification, only those features that are most discriminant will be retained for identification. This is achieved through ranking the features according to their differentiating capabilities. Lastly, since events belonging to unenrolled classes may happen in practice, an approach to isolate them before identification is required. This is also considered in this study. All the proposed algorithms will be evaluated using data generated by the Maanshan NPP simulator, Modular Accident Analysis Program Version 5 (MAAP5). Eleven event categories having 135 initiating events are included in demonstrating the efficacy and robustness of the proposed algorithms.

[1] CNSC, “Fukushima task force report,” ed: INFO-0824, Ottawa, Canada, 2011.
[2] IAEA, “Defining initiating events for purposes of probabilistic safety assessment,” IAEA-TECDOC-719, 1993.
[3] K. C. Kwon and J. H. Kim, “Accident identification in nuclear power plants using hidden Markov models,” Eng. Appl. Artif. Intell. 12, 1999, 491-501.
[4] M. G. Na, S. H. Shin, S. M. Lee, D. W. Jung, S. P. Kim, J. H. Jeong, and B. C. Lee, “Prediction of major transient scenarios for severe accidents of nuclear power plants,” IEEE Trans. Nucl. Sci. 51, 2004, 313-321.
[5] Y. G. No, J. H. Kim, M. G. Na, D. H. Lim, and K. I. Ahn, “Monitoring severe accidents using AI techniques,” Nucl. Eng. Technol. 44, 2012, 393-404.
[6] K.-C. Kwon, “HMM-based transient identification in dynamic process,” Int. J. Control Autom. Syst. 2, 2000, 40-46.
[7] J. A. C. C. Medeiros and R. Schirru, “Identification of nuclear power plant transients using the Particle Swarm Optimization algorithm,” Ann. Nucl. Energy 35, 2008, 576-582.
[8] A. dos Santos Nicolau, R. Schirru, and A. A. de Moura Meneses, “Quantum evolutionary algorithm applied to transient identification of a nuclear power plant,” Prog. Nucl. Energy 53, 2011, 86-91.
[9] M. Marseguerra, E. Zio, A. Oldrini, and E. Brega, “Fuzzy identification of transients in nuclear power plants,” Nucl. Eng. Des. 225, 2003, 285-294.
[10] M. J. Embrechts and S. Benedek, “Hybrid identification of nuclear power plant transients with artificial neural networks,” IEEE Trans. Ind. Electron. 51, 2004, 686-693.
[11] T. Santosh, G. Vinod, R. Saraf, A. Ghosh, and H. Kushwaha, “Application of artificial neural networks to nuclear power plant transient diagnosis,” Reliab. Eng. Syst. Saf. 92, 2007, 1468-1472.
[12] D. Roverso, “Soft computing tools for transient classification,” Inf. Sci. 127, 2000, 137-156.
[13] T. H. Lin, S. C. Wu, K. Y. Chen, H. P. Chou, “Feature extraction and sensor selection for NPP initiating event identification,” Ann. Nucl. Energy 103, 2017, 384-392.
[14] NRC, “Anticipated transients without scram for light water reactors,” NUREG-0460, 1978.
[15] K. Moshkbar-Bakhshayesh and M. B. Ghofrani, “Transient identification in nuclear power plants: A review,” Prog. Nucl. Energy 67, 2013, 23-32.
[16] J. Galbally and D. Galbally, “A pattern recognition approach based on DTW for automatic transient identification in nuclear power plants,” Ann. Nucl. Energy 81, 2015, 287-300.
[17] W. Cholewa, W. Frid, and M. Bednarski, “Identification of loss-of-coolant accidents in LWRs by inverse models,” Nucl. Technol. 147, 2004, 216-226.
[18] Y.-K. Liu, G.-H. Wu, C.-L. Xie, Z.-Y. Duan, M.-J. Peng, and M.-K. Li, “A fault diagnosis method based on signed directed graph and matrix for nuclear power plants,” Nucl. Eng. Des. 297, 2016, 166-174.
[19] K. Mo, S. J. Lee, and P. H. Seong, “A dynamic neural network aggregation model for transient diagnosis in nuclear power plants,” Prog. Nucl. Energy 49, 2007, 262-272.
[20] D. Roverso, “Plant diagnostics by transient classification: The aladdin approach,” Int. J. Intell. Syst. 17, 2002, 767-790.
[21] C. Gottlieb, V. Arzhanov, W. Gudowski, and N. Garis, “Feasibility study on transient identification in nuclear power plants using support vector machines,” Nucl. Technol. 155, 2006, 67-77.
[22] 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, 2008, 628-636.
[23] S. Benedek and M. Embrechts, “Rapid identification of nuclear power plant malfunctions with artificial neural networks via Fourier transformed signals,” Proc. ANNNIE, 1996.
[24] H. Asaka, Y. Kukita, T. Yonomoto, and K. Tasaka, “Results of 0.5% hot-leg break loss-of-coolant accident experiments at ROSA-IV/LSTF: the effect of break orientation,” Nuc. Technol. 96, 1991, 202-214.
[25] J. Han, J. Pei, M. Kamber, Data mining: concepts and techniques, Elsevier, 2011.
[26] R. E. Uhrig and J. W. Hines, “Computational intelligence in nuclear engineering,” Nucl. Eng. Technol. 37, 2005, 127-138.
[27] I. Attieh, A. Gribok, J. Hines, and R. Uhrig, “Pattern recognition techniques for transient detection to enhance nuclear reactor’s operational safety,” in Proc. 25th CNS/CNA Annu. Student Conf, 2000.
[28] 沸水式核能電廠，台灣電力公司第二核能發電廠，1979。
[29] 核能發電訓練基本教材-壓水式反應器系統介紹，台灣電力公司第三核能發電廠，1983。
[30] B. M. Wise, N. B. Gallagher, “The process chemometrics approach to process monitoring and fault detection,” J. Process Control 6, 1996, 329-348.
[31] L. H. Chiang, E. L. Russell, R. D. Braatz, Fault detection and diagnosis in industrial systems, Springer Science & Business Media, 2000.
[32] S. Burrus, R. Gopinath and H. Guo, Introduction to Wavelets and Wavelet Transforms: A Primer, 1/e, Englewood Cliffs, NJ: Prentice-Hall, 1998.
[33] I. Jolliffe, Principal component analysis, Wiley Online Library, 2002.
[34] 楊昭義，核能發電工程學，水牛出版社，1997。
[35] A. L. Swindlehurst, “Time delay and spatial signature estimation using known asynchronous signals,” IEEE Trans. Signal Process. 46.2, 1998, 449-462.
[36] Y. W. Chen, C. J. Lin, “Combining SVMs with various feature selection strategies,” Feature extraction. Springer, 2006, 315-324.
[37] S. Theodoridis, K. Koutroumbas, Pattern recognition (4th edition), Academic press, 2009.
[38] B. W. Silverman, Density estimation for statistics and data analysis, CRC press, 1986.
[39] M. Mohri, Foundations of machine learning, MIT press, 2012.