Fault Detection of Three-Phase Controlled Rectifiers Using Supervised Machine Learning Algorithms

doi:10.1109/ACCESS.2025.3618193

Return

Fault Detection of Three-Phase Controlled Rectifiers Using Supervised Machine Learning Algorithms

2025-01-01

DOI：10.1109/ACCESS.2025.3618193

Original

Save

View PDF

Abstract

En 中文

Power converters are integral to modern power systems and industrial applications, facilitating efficient and reliable energy transfer between sources and loads. However, their widespread use increases the likelihood of faults, particularly open-circuit faults, and performance degradation due to prolonged or harsh operating conditions. This study investigates the effectiveness of five supervised machine learning (ML) classifiers: Random Forest (RF), Support Vector Machine (SVM), Naïve Bayes (NB), K-Nearest Neighbors (KNN), and Stochastic Gradient Descent (SGD), in detecting three fault types in a three-phase controlled rectifier: open-circuit faults, wear-out faults, and their combination. The dataset was generated entirely using a detailed MATLAB/Simulink model of the rectifier, ensuring controlled fault injection for training and evaluation of the classifiers. Four key electrical features: output voltage, voltage ripple, voltage ratio, and impedance, were extracted and analyzed to assess their impact on fault detection performance. The study provides a detailed performance comparison of the classifiers based on classification accuracy and execution time. Results show that the KNN algorithm offers the best trade-off between precision and computational execution time having achieved a precision of 95.59% with an execution time of only 0.014 s while RF achieves the highest precision of 99.88% but with an execution time of 1.258 s. The study highlights the potential of ML-based approaches for accurate and timely fault detection in power converters, contributing to improved system reliability and operational safety.

Keywords:

Power electronic converter fault detection

open-circuit and wear-out fault

feature-based fault diagnosis

AI Summary

Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.