Eventos Anais de eventos
DINAME 2023
XIX International Symposium on Dynamic Problems of Mechanics
Data-driven Gear Faults Diagnostics using Bayesian Neural Networks
Submission Author:
Matheus de Moraes , SP
Co-Authors:
Matheus de Moraes, Helio Fiori de Castro
Presenter: Matheus de Moraes
doi://10.26678/ABCM.DINAME2023.DIN2023-0076
Abstract
Condition monitoring of gearboxes is a powerful technique in machinery maintenance. Artificial intelligence is nowadays supporting fault detection of several kinds of components. The objective of this research is to propose a datadriven methodology of gear faults diagnostics that accomplishes uncertainty quantification. Gear fault vibration signals obtained from an open database were converted into images that were used to train a Bayesian neural network. Test results for the Bayesian neural network fault detection capability showed high accuracy and allowed the uncertainties in the fault diagnostics to be properly quantified.
Keywords
gearboxes, condition monitoring, Bayesian neural networks
