A Stacking Ensemble Technique for Mechanical Fault Detection in Induction Motors

Abstract

The effective and accurate detection of faults in inductive electric motors is crucial to ensure their reliability and minimize costly downtime. In this study, after discussing various faults, including imbalance faults, that can arise in inductive motors for various industrial applications, we have presented a stacking ensemble learning technique for fault detection. In this technique, SVM, ELM, and MLP were considered as base learners while logistic regression was used as the meta-model. The results of the experiment have underscored the importance and effectiveness of using Machine Learning, particularly the stacking ensemble learning technique, to improve the accuracy of fault detection in inductive motors. Future work exploring and comparing other ensemble learning techniques to the presented stacking technique, and autoencoders can offer an exciting avenue to consider.