Nuclear Power Plant accident identification system with “don’t know” response capability: Novel deep learning-based approaches

Abstract When it comes to Nuclear Power Plants (NPP) operation, one of the major challenges in the Human Factors Engineering area is to provide reliable support systems, capable of rapidly identifying abnormal (transient and accident) events. Recently, Deep Learning based systems have been able to achieve promising results in this Nuclear Accident Identification Problem (NAIP). However, for these approaches to be deployed online as effective classification systems, it is necessary to supply the Deep Neural Networks with the ability to generate a “don’t know” response. This work presents, thus, novel approaches to supply a previously developed deep learning-based system for the NAIP with this indispensable feature. Multiple experiments using data of 13 operational situations simulated for a Brazilian pressurized water reactor (PWR) were designed, resulting in up to 99.88% of average performance, in terms of correct “don’t know” classifications, while retaining a substantial amount of 94.56% correct event identifications.

[1]  Christopher Leckie,et al.  High-dimensional and large-scale anomaly detection using a linear one-class SVM with deep learning , 2016, Pattern Recognit..

[2]  Yoshua Bengio,et al.  Deep Sparse Rectifier Neural Networks , 2011, AISTATS.

[3]  Khalil Moshkbar-Bakhshayesh,et al.  Transient identification in nuclear power plants: A review , 2013 .

[4]  Jürgen Schmidhuber,et al.  Deep learning in neural networks: An overview , 2014, Neural Networks.

[5]  Roberto Schirru,et al.  A new methodology for diagnosis system with ‘Don’t Know’ response for Nuclear Power Plant , 2017 .

[6]  Terrance E. Boult,et al.  Multi-class Open Set Recognition Using Probability of Inclusion , 2014, ECCV.

[7]  Riccardo Poli,et al.  Particle swarm optimization , 1995, Swarm Intelligence.

[8]  Terrance E. Boult,et al.  Probability Models for Open Set Recognition , 2014, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  D. Chakraborty,et al.  Making a multilayered perceptron network say - "don't know" when it should , 2002, Proceedings of the 9th International Conference on Neural Information Processing, 2002. ICONIP '02..

[10]  Terrance E. Boult,et al.  Towards Open World Recognition , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[11]  Randy C. Paffenroth,et al.  Anomaly Detection with Robust Deep Autoencoders , 2017, KDD.

[12]  Kee-Choon Kwon,et al.  Accident identification in nuclear power plants using hidden Markov models , 1999 .

[13]  Matheus Gutoski,et al.  A clustering-based deep autoencoder for one-class image classification , 2017, 2017 IEEE Latin American Conference on Computational Intelligence (LA-CCI).

[14]  Alexander Binder,et al.  Deep One-Class Classification , 2018, ICML.

[15]  Jie Lin,et al.  Nuclear power plant transient diagnostics using artificial neural networks that allow 'don't-know' classifications , 1995 .

[16]  Nikhil R. Pal,et al.  How to make a neural network say "Don't know" , 2018, Inf. Sci..

[17]  Roberto Schirru,et al.  A New Approach for Transient Identification with “Don’t Know” Response Using Neural Networks , 2002 .

[18]  H. Bourlard,et al.  Auto-association by multilayer perceptrons and singular value decomposition , 1988, Biological Cybernetics.

[19]  Dan Guo,et al.  Research on intelligent fault diagnosis method for nuclear power plant based on correlation analysis and deep belief network , 2018, Progress in Nuclear Energy.

[20]  Wei Xu,et al.  Improving one-class SVM for anomaly detection , 2003, Proceedings of the 2003 International Conference on Machine Learning and Cybernetics (IEEE Cat. No.03EX693).

[21]  Bernhard Schölkopf,et al.  Support Vector Method for Novelty Detection , 1999, NIPS.

[22]  Roberto Schirru,et al.  Deep rectifier neural network applied to the accident identification problem in a PWR nuclear power plant , 2019, Annals of Nuclear Energy.

[23]  Roberto Schirru,et al.  Genetic programming applied to the identification of accidents of a PWR nuclear power plant , 2019, Annals of Nuclear Energy.

[24]  Roberto Schirru,et al.  A neural model for transient identification in dynamic processes with “don't know” response , 2003 .

[25]  Mark A. Kramer,et al.  Autoassociative neural networks , 1992 .

[26]  Keun Ho Ryu,et al.  Unsupervised Novelty Detection Using Deep Autoencoders with Density Based Clustering , 2018, Applied Sciences.

[27]  Jin Jiang,et al.  Applications of fault detection and diagnosis methods in nuclear power plants: A review , 2011 .

[28]  Bo Zong,et al.  Deep Autoencoding Gaussian Mixture Model for Unsupervised Anomaly Detection , 2018, ICLR.

[29]  Thien M. Ha,et al.  The Optimum Class-Selective Rejection Rule , 1997, IEEE Trans. Pattern Anal. Mach. Intell..