Neutron Noise-Based Anomaly Classification and Localization Using Machine Learning

Abstract

A methodology is proposed in this paper allowing the classification of anomalies and subsequently their possible localization in nuclear reactor cores during operation. The method relies on the monitoring of the neutron noise recorded by in-core neutron detectors located at very few discrete locations throughout the core. In order to unfold from the detectors readings the necessary information, a 3-dimensional Convolutional Neural Network is used, with the training and validation of the network based on simulated data. In the reported work, the approach was also tested on simulated data. The simulations were carried out in the frequency domain using the CORE SIM+ diffusion-based two-group core simulator. The different scenarios correspond to the following cases: a generic “absorber of variable strength”, axially travelling perturbations at the velocity of the coolant flow (due to e.g. fluctuations of the coolant temperature at the inlet of the core), fuel assembly vibrations, control rod vibrations, and core barrel vibrations. In all those cases, various frequencies were considered and, when relevant, different locations of the perturbations and different vibration modes were taken into account. The machine learning approach was able to correctly identify the different scenarios with a maximum error of 0.11%. Moreover, the error in localizing anomalies had a mean squared error of 0.3072 in mesh size, corresponding to less than 4 cm. The proposed methodology was also demonstrated to be insensitive to parasitic noise and will be tested on actual plant data in the near future.

Localizing Perturbations in Pressurized Water Reactors Using One-Dimensional Deep Convolutional Neural Networks

“…In recent years, the development of massive parallel processing computation systems at reduced cost (e.g., in the from of graphical or tensor processing units) has permitted the training of much larger ANN architectures on large volumes of data, leading to the introduction of deep learning approaches in NPP operation and safety. In this respect and in the framework of the CORTEX project, a three-dimensional (3D) convolutional neural network (CNN) model in the frequency domain was adapted to the localization problem [ 7 ]. Other works analyzed neutron noise signals using recurrent neural networks (RNNs) [ 8 ] and long short-term memory (LSTM) units [ 8 , 9 ].…”

Characterization of High Harmonic Frequencies in Reactor Noise Experiments Within the CORTEX Project

“…The aim of the EU H2020 CORTEX project [6] is to develop tools and techniques for the localization and identification of noise sources by neutron flux measurements using multiple detectors. This is performed by analyzing the measured data by machine learning algorithms [7], [8], which have been previously trained using validated neutron noise propagation simulation computer codes [9]- [15]. The code validation is also performed within the scope of this project by performing experiments at zero power reactor facilities capable of inducing neutron noise: the Techniche Universität Dresden (TUD) Ausbildungskernreactor 2 (AKR-2) [16] equipped with a rotating and oscillating neutron absorber, entering the core from the side through its experimental channels (Figure 1) and the École Polytechnique Fédérale de Lausanne (EPFL) CROCUS reactor [17] with CROCUS Oscillator for Lateral Increase Between U -metal Rods and Inner zone (COLIBRI) experiment [18] located inside the reactor pool on the west side, oscillating up to 18 fuel rods and Pile Oscillator for Localized and Low Effect Noise (POLLEN) oscillating neutron absorber [19] usually installed at the core center.…”