EUVIMEDEUVIMED entdecken
Crossref

Hybrid Attention and Transformer Network for Fault Signal Analysis in Rotating Machinery

Depei Shao, Dezun Zhao, Tianyang Wang

IEEE Transactions on Instrumentation and Measurement · 2026 · 75 · 6508709-6508709

DOI: 10.1109/tim.2026.3687356 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Instrumentation and Measurement, 2026. Verfasst von Depei Shao und 2 weiteren Autor:innen. Die Arbeit nennt 24 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Fault signals of rotating machinery exhibit pronounced nonstationary behavior under time-varying conditions, posing significant challenges for current state-of-the-art deep learning-based time–frequency analysis (TFA) methods, which often fail to achieve high-resolution characterization of closely spaced and crossover instantaneous frequencies (IFs). To address this challenge, a novel TFA model, termed the hybrid attention and Transformer network (HATN), is proposed. In the HATN, the multistage hybrid enhancement module (HEM) and adaptive layer attention module (ALAM) are proposed to optimize the time–frequency representation (TFR). In particular, an efficient grouped Swin Transformer (EGST) is first designed in the HEM to efficiently capture the time and frequency transformation characteristics of the TFR by exploiting global dependency. Second, the EGST and two attention mechanisms are alternately arranged to recalibrate the importance of dynamic TFR, and the energy concentration is iteratively enhanced. Finally, the ALAM is developed to adaptively adjust the hierarchical weight based on cross-layer feature correlation to strengthen the transfer and fusion of key features. The simulated signal, bearing signal with multiple faults, and planetary gearbox signal with a single fault are used to analyze, and the effectiveness of the HATN is verified by comparing with the various TFA methods.

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Shao, D., Zhao, D., & Wang, T. (2026). Hybrid Attention and Transformer Network for Fault Signal Analysis in Rotating Machinery. IEEE Transactions on Instrumentation and Measurement, 75, 6508709-6508709. https://doi.org/10.1109/tim.2026.3687356

Vancouver

Shao D, Zhao D, Wang T. Hybrid Attention and Transformer Network for Fault Signal Analysis in Rotating Machinery. IEEE Transactions on Instrumentation and Measurement. 2026;75:6508709-6508709. doi:10.1109/tim.2026.3687356

BibTeX
@article{Shao2026,
  author = {Shao, Depei and Zhao, Dezun and Wang, Tianyang},
  title = {Hybrid Attention and Transformer Network for Fault Signal Analysis in Rotating Machinery},
  journal = {IEEE Transactions on Instrumentation and Measurement},
  year = {2026},
  volume = {75},
  pages = {6508709--6508709},
  doi = {10.1109/tim.2026.3687356}
}
Kostenlos starten →