EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

A Noise‐Robust Multi‐Scale Dilated Convolutional Neural Network With Scale Attention for Bearing Fault Diagnosis

Chun‐Yao Lee, Guang‐Lin Zhuo, Tzu‐Hsien Chuang, Chung‐Hao Huang, Hong‐Chan Chang, Kuan‐Yu Huang, Truong‐An Le

IET Electric Power Applications · 2026-01 · 20 (1)

DOI: 10.1049/elp2.70269 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IET Electric Power Applications, online erschienen am 17.11.2026. Verfasst von Chun‐Yao Lee und 6 weiteren Autor:innen. Die Arbeit nennt 54 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: National Science and Technology Council.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This study proposes a noise‐robust one‐dimensional convolutional neural network (1D‐CNN) for rolling bearing fault diagnosis under low signal‐to‐noise ratio (low‐SNR) conditions. The proposed architecture integrates a large‐kernel stem, dilated multi‐scale branches, and a scale attention mechanism to enhance multi‐scale feature learning and adaptive feature fusion. Experimental results on the Case Western Reserve University (CWRU) and Paderborn University (PU) datasets demonstrate the effectiveness of the proposed method. The proposed model achieves classification accuracies of 96.8% and 91.9% under −9 and −2 dB conditions, respectively, representing improvements of 18.7 and 13.4 percentage points over the baseline MSCNN model. In addition, the proposed architecture requires only 82.85 M FLOPs and 0.13 M parameters. Compared with MSCNN, the proposed model substantially reduces the parameter count while maintaining superior diagnostic performance with only a modest increase in computational cost. These results demonstrate that the proposed architecture achieves a favourable accuracy–efficiency trade‐off and strong noise robustness for rolling bearing fault diagnosis under the evaluated AWGN conditions.

Förderung

  • National Science and Technology Council

Zitieren

APA

Lee, C., Zhuo, G., Chuang, T., Huang, C., Chang, H., Huang, K., & Le, T. (2026). A Noise‐Robust Multi‐Scale Dilated Convolutional Neural Network With Scale Attention for Bearing Fault Diagnosis. IET Electric Power Applications, 20(1). https://doi.org/10.1049/elp2.70269

Vancouver

Lee C, Zhuo G, Chuang T, Huang C, Chang H, Huang K, et al. A Noise‐Robust Multi‐Scale Dilated Convolutional Neural Network With Scale Attention for Bearing Fault Diagnosis. IET Electric Power Applications. 2026;20(1). doi:10.1049/elp2.70269

BibTeX
@article{Lee2026,
  author = {Lee, Chun‐Yao and Zhuo, Guang‐Lin and Chuang, Tzu‐Hsien and Huang, Chung‐Hao and Chang, Hong‐Chan and Huang, Kuan‐Yu and Le, Truong‐An},
  title = {A Noise‐Robust Multi‐Scale Dilated Convolutional Neural Network With Scale Attention for Bearing Fault Diagnosis},
  journal = {IET Electric Power Applications},
  year = {2026},
  volume = {20},
  number = {1},
  doi = {10.1049/elp2.70269}
}
Kostenlos starten →