Acoustic fault diagnosis of traction motor bearing based on fusion feature
The rolling bearing is an important part of the traction motor of high-speed train, and its failure seriously affects the safety of train operation. The acoustic diagnosis method for bearing fault has the advantages of non-invasive installation and low operation and maintenance cost, but it also has...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | zho |
| Published: |
Editorial Department of Electric Drive for Locomotives
2023-03-01
|
| Series: | 机车电传动 |
| Subjects: | |
| Online Access: | http://edl.csrzic.com/thesisDetails#10.13890/j.issn.1000-128X.2023.02.012 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The rolling bearing is an important part of the traction motor of high-speed train, and its failure seriously affects the safety of train operation. The acoustic diagnosis method for bearing fault has the advantages of non-invasive installation and low operation and maintenance cost, but it also has the disadvantages of low signal-to-noise ratio and difficult to extract fault features. Machine learning has the robustness to avoid the influence of noise. Aiming at the problem that a small number of features cannot fully characterize the bearing fault when applying machine learning to acoustic fault diagnosis, this paper proposes to superimpose and fuse the Gramian angular field (GAF) and wavelet time-frequency figure to form a six channel fusion feature map to effectively characterize the bearing fault. Firstly, the traction motor bearing acoustic fault test-bed was established to obtain the fault acoustic signal. Secondly, the acoustic signal fusion feature map based on GAF was established. Then, the residual networks (ResNETs) model was used to train and verify the fault classification model for the features of the fusion feature map, and the accuracy was compared with the fault classification method with a single feature map as the feature. The results show that the acoustic fault classification model based on GAF fusion feature map has an accuracy of 99.89%, so the fusion feature map can more effectively reflect the bearing fault. |
|---|---|
| ISSN: | 1000-128X |