EUVIMEDEUVIMED entdecken
Crossref

Rolling Bearing Fault Diagnosis Methods Based on Adaptive Fourier–Bessel Networks

Weiwen Tian, Yanfei Liu, Yanfeng Peng, Jie Zhou, Haidong Shao, Xiaolan Hu

IEEE Sensors Journal · 01.05.2026 · 26 (9) · 13467-13481

DOI: 10.1109/jsen.2026.3673167 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.05.2026. Verfasst von Weiwen Tian und 5 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Hunan Provincial Science and Technology Innovation Talent Project, Natural Science Foundation of Hunan Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning has been widely applied in the field of fault diagnosis, but most research still focuses on time-domain feature analysis. Frequency-domain-based methods are relatively rare and often rely on the fast Fourier transform (FFT). However, traditional FFT methods suffer from limitations such as low-frequency resolution and inadequate representation capabilities for nonstationary signals, which restrict their diagnostic performance under complex operating conditions and high-noise environments. To address this, the adaptive Fourier–Bessel network (AFBNet) method has been proposed. This method leverages the high-resolution advantage of the Fourier–Bessel series (FBS) in signal processing. To address the nonstationary characteristics of fault features in rolling bearing vibration signals under varying operating conditions, an adaptive Bessel order determination (BOD) algorithm is proposed. By quantifying the time–frequency energy concentration of the signal, the algorithm adaptively adjusts the order of the Bessel function, enhancing its diagnostic stability under multiple variable operating conditions. Building upon this foundation, to address the susceptibility of signal decomposition network methods to noise interference, a band segmentation algorithm based on a joint feature loss function is proposed. By constructing a joint feature-based loss function constraint and optimally reconstructing frequency bands containing critical fault information, this approach amplifies weak fault features, thereby enhancing the model’s feature extraction capability and diagnostic accuracy. Experimental results demonstrate that AFBNet exhibits good diagnostic performance on both rolling bearing experimental data and public datasets, particularly showing high robustness and accuracy in high-noise conditions.

Förderung

  • National Natural Science Foundation of China
  • Hunan Provincial Science and Technology Innovation Talent Project
  • Natural Science Foundation of Hunan Province

Zitieren

APA

Tian, W., Liu, Y., Peng, Y., Zhou, J., Shao, H., & Hu, X. (2026). Rolling Bearing Fault Diagnosis Methods Based on Adaptive Fourier–Bessel Networks. IEEE Sensors Journal, 26(9), 13467-13481. https://doi.org/10.1109/jsen.2026.3673167

Vancouver

Tian W, Liu Y, Peng Y, Zhou J, Shao H, Hu X. Rolling Bearing Fault Diagnosis Methods Based on Adaptive Fourier–Bessel Networks. IEEE Sensors Journal. 2026;26(9):13467-13481. doi:10.1109/jsen.2026.3673167

BibTeX
@article{Tian2026,
  author = {Tian, Weiwen and Liu, Yanfei and Peng, Yanfeng and Zhou, Jie and Shao, Haidong and Hu, Xiaolan},
  title = {Rolling Bearing Fault Diagnosis Methods Based on Adaptive Fourier–Bessel Networks},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {9},
  pages = {13467--13481},
  doi = {10.1109/jsen.2026.3673167}
}
Kostenlos starten →