3D convolutional and recurrent neural networks for reactor perturbation unfolding and anomaly detection

Aiden Durrant; Georgios Leontidis; Stefanos Kollias

doi:10.1051/epjn/2019047

3D convolutional and recurrent neural networks for reactor perturbation unfolding and anomaly detection

Aiden Durrant^* (Corresponding Author), Georgios Leontidis, Stefanos Kollias

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

6 Downloads (Pure)

Abstract

With Europe's ageing fleet of nuclear reactors operating closer to their safety limits, the monitoring of such reactors through complex models has become of great interest to maintain a high level of availability and safety. Therefore, we propose an extended Deep Learning framework as part of the CORTEX Horizon 2020 EU project for the unfolding of reactor transfer functions from induced neutron noise sources. The unfolding allows for the identification and localisation of reactor core perturbation sources from neutron detector readings in Pressurised Water Reactors. A 3D Convolutional Neural Network (3D-CNN) and Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) have been presented, each to study the signals presented in frequency and time domain respectively. The proposed approach achieves state-of-the-art results with the classification of perturbation type in the frequency domain reaching 99.89% accuracy and localisation of the classified perturbation source being regressed to 0.2902 Mean Absolute Error (MAE).

Original language	English
Article number	20
Number of pages	9
Journal	EPJ Nuclear Sciences & Technologies
Volume	5
Early online date	29 Nov 2019
DOIs	https://doi.org/10.1051/epjn/2019047
Publication status	Published - 2019

Keywords

Machine Learning
Deep Learning
nuclear reactors
signal processing

Access to Document

10.1051/epjn/2019047Licence: CC BY

Durrant_EPJN_3DConvolutional_VoR
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Final published version, 1.02 MBLicence: CC BY

Cite this

@article{760f1c40554947a3a2ced73834a1c13f,

title = "3D convolutional and recurrent neural networks for reactor perturbation unfolding and anomaly detection",

abstract = "With Europe's ageing fleet of nuclear reactors operating closer to their safety limits, the monitoring of such reactors through complex models has become of great interest to maintain a high level of availability and safety. Therefore, we propose an extended Deep Learning framework as part of the CORTEX Horizon 2020 EU project for the unfolding of reactor transfer functions from induced neutron noise sources. The unfolding allows for the identification and localisation of reactor core perturbation sources from neutron detector readings in Pressurised Water Reactors. A 3D Convolutional Neural Network (3D-CNN) and Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) have been presented, each to study the signals presented in frequency and time domain respectively. The proposed approach achieves state-of-the-art results with the classification of perturbation type in the frequency domain reaching 99.89% accuracy and localisation of the classified perturbation source being regressed to 0.2902 Mean Absolute Error (MAE).",

keywords = "Machine Learning, Deep Learning, nuclear reactors, signal processing",

author = "Aiden Durrant and Georgios Leontidis and Stefanos Kollias",

note = "The research conducted was made possible through funding from the Euratom research and training programme 2014-2018 under grant agreement No 754316 for the {\textquoteleft}CORe Monitoring Techniques And EXperimental Validation And Demonstration (CORTEX){\textquoteright} Horizon 2020 project, 2017-2021. We would like to thank the Chalmers University of Technology, particularly Dr C. Demaziere, Dr P. Vinai, Dr A. Milonakis and the Paul Scherrer Institute, particularly Dr A. Dokhane and Dr V. Verma for providing the frequency and domain data respectively, for assisting us with their understanding and for collaborating with us in the analysis process.",

year = "2019",

doi = "10.1051/epjn/2019047",

language = "English",

volume = "5",

journal = "EPJ Nuclear Sciences & Technologies",

}

TY - JOUR

T1 - 3D convolutional and recurrent neural networks for reactor perturbation unfolding and anomaly detection

AU - Durrant, Aiden

AU - Leontidis, Georgios

AU - Kollias, Stefanos

N1 - The research conducted was made possible through funding from the Euratom research and training programme 2014-2018 under grant agreement No 754316 for the ‘CORe Monitoring Techniques And EXperimental Validation And Demonstration (CORTEX)’ Horizon 2020 project, 2017-2021. We would like to thank the Chalmers University of Technology, particularly Dr C. Demaziere, Dr P. Vinai, Dr A. Milonakis and the Paul Scherrer Institute, particularly Dr A. Dokhane and Dr V. Verma for providing the frequency and domain data respectively, for assisting us with their understanding and for collaborating with us in the analysis process.

PY - 2019

Y1 - 2019

N2 - With Europe's ageing fleet of nuclear reactors operating closer to their safety limits, the monitoring of such reactors through complex models has become of great interest to maintain a high level of availability and safety. Therefore, we propose an extended Deep Learning framework as part of the CORTEX Horizon 2020 EU project for the unfolding of reactor transfer functions from induced neutron noise sources. The unfolding allows for the identification and localisation of reactor core perturbation sources from neutron detector readings in Pressurised Water Reactors. A 3D Convolutional Neural Network (3D-CNN) and Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) have been presented, each to study the signals presented in frequency and time domain respectively. The proposed approach achieves state-of-the-art results with the classification of perturbation type in the frequency domain reaching 99.89% accuracy and localisation of the classified perturbation source being regressed to 0.2902 Mean Absolute Error (MAE).

AB - With Europe's ageing fleet of nuclear reactors operating closer to their safety limits, the monitoring of such reactors through complex models has become of great interest to maintain a high level of availability and safety. Therefore, we propose an extended Deep Learning framework as part of the CORTEX Horizon 2020 EU project for the unfolding of reactor transfer functions from induced neutron noise sources. The unfolding allows for the identification and localisation of reactor core perturbation sources from neutron detector readings in Pressurised Water Reactors. A 3D Convolutional Neural Network (3D-CNN) and Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN) have been presented, each to study the signals presented in frequency and time domain respectively. The proposed approach achieves state-of-the-art results with the classification of perturbation type in the frequency domain reaching 99.89% accuracy and localisation of the classified perturbation source being regressed to 0.2902 Mean Absolute Error (MAE).

KW - Machine Learning

KW - Deep Learning

KW - nuclear reactors

KW - signal processing

U2 - 10.1051/epjn/2019047

DO - 10.1051/epjn/2019047

M3 - Article

VL - 5

JO - EPJ Nuclear Sciences & Technologies

JF - EPJ Nuclear Sciences & Technologies

M1 - 20

ER -

3D convolutional and recurrent neural networks for reactor perturbation unfolding and anomaly detection

Abstract

Keywords

Access to Document

Fingerprint

Cite this