Diagnóstico de falhas em rolamentos de motores de indução trifásicos via abordagem densidade de máximos a partir do sinal da corrente do estator

Abstract

Three-phase induction motors (TIMs) are widely used in the industry due to their reliability, robustness, and low cost, being the main driving force in the industrial sector. Bearing failures account for more than 40% of induction motor failures, resulting in unexpected downtimes, increased production costs, and potential accidents. In this context, this work proposes a new non-invasive method for diagnosing bearing failures in TIMs, using stator current analysis combined with the SAC-DM (Signal Analysis based on Chaos using Density of Maxima) technique and a Multilayer Perceptron (MLP) artificial neural network.For comparison, two additional methods were implemented that use mechanical vibration and sound signals emitted by the TIM, together with the Hilbert envelope technique. The signals were collected simultaneously under six load conditions: 0%, 20%, 40%, 60%, 80%, and 100% of the nominal load. Conditions of healthy bearings, outer race failure, inner race failure, and mixed failure were analyzed. For outer race failure, the vibration signal analysis showed a maximum relative percentage error of 0.60%, while for sound the error was 0.53%. For inner race failure, the maximum relative percentage error was 0.49% for vibration and 0.36% for sound. The proposed new method, using SAC-DM, detected bearing failures with a classification accuracy of 96.82%, even with the motor operating under variable load. This new methodology can, therefore, significantly contribute to the diagnosis of bearing failures in TIMs, offering a more efficient and noninvasive alternative compared to traditional methods.