EUVIMEDEUVIMED entdecken
Crossref

Cross-Domain Fault Diagnosis of Aero-Engines Under Imbalanced Source and Few-Shot Conditions

Ying Zhao, Hao Wang, Changyi Xu, Chao Zhang

IEEE Sensors Journal · 01.06.2026 · 26 (11) · 17414-17423

DOI: 10.1109/jsen.2026.3684737 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Sensors Journal, erschienen am 01.06.2026. Verfasst von Ying Zhao und 3 weiteren Autor:innen. Die Arbeit nennt 38 Literaturangaben. Gefördert von: Natural Science Foundation of Liaoning Province of China, National Natural Science Foundation of China, Zhejiang Pioneer Plan.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Intelligent fault diagnosis of aero-engines is persistently challenged by the scarcity of fault samples and severe class distribution imbalance. To address these challenges, this article proposes a novel fault diagnosis method termed the cross-domain few-shot fault diagnosis under imbalanced source (CDFI). By leveraging a meticulously designed joint learning objective, this method effectively overcomes performance degradation caused by data scarcity, class imbalance, and domain shift. Specifically, an adaptive weighting mechanism (AWM) is proposed to dynamically adjust class weights based on the learning state, effectively mitigating bias from dominant classes to ensure minority features are fully mined. In addition, a dual-objective contrastive learning (DOCL) loss is introduced to collaboratively facilitate intra-domain structure preservation and crossdomain alignment, yielding robust discriminative and domain-invariant features. Furthermore, a random composite data augmentation (RCDA) strategy is employed to augment scarce target samples via multidimensional signal synthesis, thereby enhancing data diversity and training stability. Validations on both simulated aero-engine and real-world physical datasets confirm that CDFI outperforms advanced baselines, maintaining high diagnostic accuracy on hard-to-classify samples under domain shift.

Förderung

  • Natural Science Foundation of Liaoning Province of China
  • National Natural Science Foundation of China
  • Zhejiang Pioneer Plan

Zitieren

APA

Zhao, Y., Wang, H., Xu, C., & Zhang, C. (2026). Cross-Domain Fault Diagnosis of Aero-Engines Under Imbalanced Source and Few-Shot Conditions. IEEE Sensors Journal, 26(11), 17414-17423. https://doi.org/10.1109/jsen.2026.3684737

Vancouver

Zhao Y, Wang H, Xu C, Zhang C. Cross-Domain Fault Diagnosis of Aero-Engines Under Imbalanced Source and Few-Shot Conditions. IEEE Sensors Journal. 2026;26(11):17414-17423. doi:10.1109/jsen.2026.3684737

BibTeX
@article{Zhao2026,
  author = {Zhao, Ying and Wang, Hao and Xu, Changyi and Zhang, Chao},
  title = {Cross-Domain Fault Diagnosis of Aero-Engines Under Imbalanced Source and Few-Shot Conditions},
  journal = {IEEE Sensors Journal},
  year = {2026},
  volume = {26},
  number = {11},
  pages = {17414--17423},
  doi = {10.1109/jsen.2026.3684737}
}
Kostenlos starten →