Detecção de falhas de estator em motores de indução trifásicos através de análise acústica

Abstract

Three-phase induction motors are widely used due to their characteristics, such as simple construction, easy maintenance, high adaptability and high cost-effectiveness. However, these motors are often subject to different types of stress which, over time, can lead to failures. Among the most recurrent faults is the short-circuit in the stator, which is the focus of this research. The aim of the study is to detect and classify this fault using audio signals acquired with the motor in normal and fault conditions, using various classification models, such as k-Nearest Neighbour, Random Forest, Decision Tree, Multilayer Perceptron, Support Vector Machines and Extreme Gradient Boosting. The models tested showed high performance, with accuracies of up to 100% or results very close to this value, guaranteeing efficient detection of the short circuit in the stator. These results indicate that audio analysis combined with machine learning techniques could be a promising approach for monitoring motor faults, contributing to predictive maintenance and reducing operating costs.