A Novel Small-Sample Fault Diagnosis Method for Rolling Bearings via Continuous Wavelet Transform and Siamese Neural Network
IEEE Sensors Journal · 01.08.2024 · 24 (15) · 24988-24996
DOI: 10.1109/jsen.2024.3409768 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.08.2024. Verfasst von Xufeng Zhao und 4 weiteren Autor:innen. Die Arbeit wurde bisher 36-mal zitiert (Crossref) und nennt 29 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Aeronautical Science Foundation of China Aviation Academy, Postgraduate Research and Practice Innovation Program of Jiangsu Province.
Abstract (Originalsprache)
Fault diagnosis is essential to guarantee the reliability and safety of rolling bearings. However, the current research mainly focuses on a large amount of data and neglects the problem of insufficient sample size of rolling bearings in actual engineering. Therefore, this study investigates a novel small-sample fault diagnosis method for rolling bearings based on continuous wavelet transform and Siamese neural network (CWT-SNN). First, each raw signal captured by the rolling bearings is segmented into samples by sliding window and slicing operations and converted into signal time–frequency maps by CWT. Second, a 2-D convolutional neural network (2DCNN) is constructed to extract features from all fault samples and generate embedding vectors. Finally, sample pairs are constructed to input the embedding vectors into the Siamese neural network (SNN) and measure their similarity, thus realizing rolling bearing fault diagnosis under small-sample conditions. Experimental results show that the diagnostic accuracy of CWT-SNN model is improved by more than 10% compared with other models when there are only 60 fault samples. Meanwhile, CWT-SNN diagnostic accuracy can reach more than 99% with sample sizes larger than 120 and shows better generalization performance in new categories and new operating conditions.
Förderung
- National Natural Science Foundation of China
- Aeronautical Science Foundation of China Aviation Academy
- Postgraduate Research and Practice Innovation Program of Jiangsu Province
Zitieren
APA
Zhao, X., Wang, L., Yang, M., Chen, Y., & Xiang, J. (2024). A Novel Small-Sample Fault Diagnosis Method for Rolling Bearings via Continuous Wavelet Transform and Siamese Neural Network. IEEE Sensors Journal, 24(15), 24988-24996. https://doi.org/10.1109/jsen.2024.3409768
Vancouver
Zhao X, Wang L, Yang M, Chen Y, Xiang J. A Novel Small-Sample Fault Diagnosis Method for Rolling Bearings via Continuous Wavelet Transform and Siamese Neural Network. IEEE Sensors Journal. 2024;24(15):24988-24996. doi:10.1109/jsen.2024.3409768
BibTeX
@article{Zhao2024,
author = {Zhao, Xufeng and Wang, Lubing and Yang, Mengshu and Chen, Ying and Xiang, Jiawei},
title = {A Novel Small-Sample Fault Diagnosis Method for Rolling Bearings via Continuous Wavelet Transform and Siamese Neural Network},
journal = {IEEE Sensors Journal},
year = {2024},
volume = {24},
number = {15},
pages = {24988--24996},
doi = {10.1109/jsen.2024.3409768}
}
EUVIMED