Machine Learning-Based Fault Diagnosis for a PWR Nuclear Power Plant

Abstract

Fault detection and diagnosis (FDD) systems can reduce high costs and energy consumption. This paper presents a machine learning-based fault detection and diagnosis (FDD) technique for actuators and sensors in a pressurized water reactor (PWR). In the proposed FDD framework, faults are first detected using a shallow neural network. Second, fault diagnosis is performed using 15 different classifiers provided in the MATLAB Classification Learner toolbox, including support vector machine (SVM), Knearest neighbor (KNN), and ensemble. Several classifiers were found to provide superior classification performance, including medium KNN, cubic KNN, cosine KNN, weighted KNN, fine Gaussian SVM, quadratic SVM, medium Gaussian SVM, coarse Gaussian, bagged trees, and subspace KNN. The accuracy of the FDD approach was demonstrated using a set of simulation results. INDEX TERMS Fault detection and diagnosis (FDD), nuclear reactor, machine learning, neural networkThis article has been accepted for publication in IEEE Access.

Fault States Diagnosis of Marine Diesel Engine Valve Based on a Modified VGG16 Transfer Learning Method

“…In this study, the initial learning rate of cosine annealing is 10 −5 , and the minimum value of the learning rate is set as 10 −6 . Equation (11) is the formula of learning rate with the number of iterations, where i denotes the number of restarts, η i min and η i max are the range of learning rate, T cur denotes the currently executed epochs, and T i denotes the total number of epochs in the ith run…”

“…Te second category is the traditional machine learning algorithm, such as stream learning, support vector machines (SVM), random forests, and back propagation neural networks (BPNN), which can simplify the signal processing process and improve the efciency of identifcation compared with the traditional vibration signal analysis method [11][12][13][14]. Tese diagnostic methods can achieve high diagnostic accuracy for diesel engine fault datasets due to their efective feature extraction and reasonable classifer design.…”

A Novel Reduced-Order Analytical Fault Diagnosis Model for Power Grid

“…Moreover, Refs. [14][15] adopted a methodology for integrating multiple artificial neural networks or combining artificial neural networks with other algorithms, such as the elgamal encryption algorithm, to further improve the accuracy of the fault analysis results.…”

An overview of power reactor kinetics and control in load-following operation modes

“…While the majority of the publications on PWR control methods is of academic nature, some experimental studies report practical experience with load-following operation of nuclear power plants or discuss the control methods used in practice (Sipush et al, 1976;Meyer et al, 1978;Onoue et al, 2003;Franceschini and Petrovic, 2008;Wei and Zhao, 2015;Zhang et al, 2015;Lee et al, 2020;Park et al, 2022). In the last decade, modern deeplearning artificial intelligence approaches are gaining traction also in related fault-detection (Hu et al, 2021;Naimi et al, 2022a;Kollias et al, 2022), reactor design (Dzianisau et al, 2022), fuel management (Hassan et al, 2021;Che et al, 2022), and safety analysis (Demazière, Christophe et al, 2020;Gomez-Fernandez et al, 2020;Ayodeji et al, 2022;Racheal et al, 2022). The integration with renewable energy sources and the energy storage systems are additional important emerging research fields (Denholm et al, 2012;Borowiec et al, 2019;Bragg-Sitton et al, 2020;Kim and Alameri, 2020).…”