Eventos Anais de eventos
DINAME 2023
XIX International Symposium on Dynamic Problems of Mechanics
Evaluation of a methodology for low-speed bearing fault diagnosis
Submission Author:
Thiago Barroso Costa , PA
Co-Authors:
Thiago Barroso Costa, João Lucas Lobato Soares, Walter dos Santos Sousa, Alexandre Mesquita, André Luiz Amarante Mesquita, Elton Prestes de Souza , Danilo Braga
Presenter: Thiago Barroso Costa
doi://10.26678/ABCM.DINAME2023.DIN2023-0143
Abstract
Rolling element bearings are critical components because they often represent high incidence of rotating machinery failures. In the last decades, some researchers have proposed methods of monitoring, detection, diagnosis and prognosis based on artificial intelligence tools, seeking for more reliable decision-making in predictive maintenance. In this sense, application of machine learning has been investigated for fault diagnosis of low-speed bearings due to the monitoring of their condition be more challenger than high-speed bearings, and traditional methods become limited. Hence, this work proposes evaluate the MLE (Maximal Lyapunov Exponent) and the Hjorth’s parameters as features extracted from the vibration signals measured in a test rig at 60 rpm with different frequency range and sampling rate. First, those features were ranked according to their significance estimated by Welch’s t-test, i.e., the capability to maximize the class separation, and they also were compared to others time domain features. Secondly, the root mean square and standard deviation presented high separability and were associated to Hjorth’s parameters and MLE, generating a matrix of six features. Finally, SMV algorithm was applied and the results showed the ensemble six predictors was a good approach chosen for discriminate the classes, even with PCA dimensionality reduction.
Keywords
Vibration Analysis, Low-speed bearing, Fault Diagnosis, MLE, Hjorth’s parameters, Welch’s t-test, svm
