On the Use of Convolutional Neural Network for Bearing Fault Diagnosis with Small Data
Rolling element bearings are commonly used in supporting rotor components and assemblies in rotating machinery. Bearing defects can lead to undesirable vibrations, noise, or machine failure. Modern predictive maintenance techniques are increasingly adopting artificial intelligence (AI) techniques for bearing fault diagnosis. In this work, wavelet analysis is used to process vibration signals from three bearing cases. These are: no fault, inner race fault, and ball fault under varying rotating speed. Small set of the wavelet scalogram images are fed into a Convolutional Neural Network (CNN) model for the classification of bearing defects into four classes based on operating speeds. Even with small sample size for training, we have been to achieve classification accuracies of around 90% with a specific combination of CNN parameters.
Files
Metadata
| Work Title | On the Use of Convolutional Neural Network for Bearing Fault Diagnosis with Small Data |
|---|---|
| Access | |
| Creators |
|
| Keyword |
|
| License | In Copyright (Rights Reserved) |
| Work Type | Article |
| Publisher |
|
| Publication Date | May 18, 2023 |
| Publisher Identifier (DOI) |
|
| Related URLs | |
| Deposited | February 18, 2025 |
Versions
Analytics
Collections
This resource is currently not in any collection.