DIAGNOSIS AND CONDITION MONITORING OF ELECTRICAL MACHINES USING MACHINE LEARNING TECHNIQUES

Abstract

The simplest components of the electrical industry and power systems are electrical machines such as motors, generators, and transformers, which can lead to life-threatening interference and even economic damage when they are not planned to be used. To address this challenge, machine learning (ML) methods have emerged as effective diagnostics methods of fault diagnosis, and condition monitoring. It is in this paper that systematic analysis will be provided regarding the use of ML algorithms in the process of identifying, classifying, and predicting machine faults through online vibration, temperature, and current measurements. The proposed methodology takes into consideration the endogenous data pre-processing, feature extraction and supervised learning models such as Support Vector Machine (SVM), Random Forest (RF) and Artificial Neural Networks (ANN) to identify faults efficiently. But realistic conditions like information asymmetry, limit to sensor location and computational expenses in real time are at stake. Specifically, the future research directions involve the combination of edge-AI structures, federation learning of distributed systems and federation of hybrid deep learning models to provide scaled, adaptive and energy efficient fault diagnosis to conceive Industry 5.0.