EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Measurement-Efficient Few-Shot Vibration Fault Diagnosis via Physics-Informed Self-Supervised Learning and Adaptive Early Stopping

Zongzhe Ni iD, Xiancheng Ji iD, Jianjun Yi iD, Nuozhou Li iD, Hongxing Wang iD, Yifan Liu iD, Ying Yan iD

Sensors · 04.07.2026 · 26 (13) · 4252

DOI: 10.3390/s26134252 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 04.07.2026. Verfasst von Zongzhe Ni und 6 weiteren Autor:innen. Beteiligte Einrichtung: School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai 200237, China. Die Arbeit nennt 62 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Shanghai Science and Technology Action Plan, Shanghai Aerospace Science and Technology Innovation Fund, the Special Fund Technology Innovation Support Project of Shanghai.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Vibration-based fault diagnosis is widely used for rotating machinery health monitoring, but practical diagnosis is often limited by scarce fault labels and uncertain measurement length. Longer vibration records can improve decision reliability but increase sensing and computational cost, whereas overly short records may yield unreliable predictions under noise and measurement corruptions. This paper studies few-shot fault diagnosis as a measurement-constrained decision task, in which the model identifies the fault class and determines when sufficient vibration evidence has been acquired. We propose a measurement-efficient diagnosis framework that combines prior knowledge from unlabeled healthy signals, physically constrained augmentation of scarce labeled samples, and adaptive early stopping in a shared one-dimensional feature extractor. The framework is evaluated on the UORED-VAFCLS and Paderborn University bearing datasets under 6-, 8-, and 10-shot settings with controlled corruption levels. Results show robust diagnostic performance with fewer acquired vibration windows than with fixed-length inference. In the representative PU-Hard 8-shot setting, the proposed method achieves 80.26% accuracy with an average of 1.2432 acquired windows and reduces the evaluation cost J from 0.3929 to 0.2596 compared with fixed four-window inference. These results indicate that adaptive measurement improves the accuracy–cost trade-off in few-shot vibration diagnosis.

Schlagwörter

Stichwörter der Autor:innen

data augmentationdeep reinforcement learningfew-shot fault diagnosisself-supervised learning

Autor:innen und Institutionen

  • School of Mechanical and Power Engineering, East China University of Science and Technology, Shanghai 200237, China.
  • Robotics Engineering Center, The 21st Research Institute of CETC, Shanghai 200233, China.

ORCID-Profile: Zongzhe Ni, Xiancheng Ji, Jianjun Yi, Nuozhou Li, Hongxing Wang, Yifan Liu, Ying Yan

Förderung

  • Shanghai Science and Technology Action Plan
  • Shanghai Aerospace Science and Technology Innovation Fund
  • the Special Fund Technology Innovation Support Project of Shanghai
  • the National Defense Basic Scientific Research Program of China
  • the ``Pioneer'' and ``Leading Goose'' R\&D Program of Zhejiang Province
  • Special Fund Technology Innovation Support Project of Shanghai
  • National Defense Basic Scientific Research Program of China
  • “Pioneer” and “Leading Goose” R&D Program of Zhejiang Province

Zitieren

APA

Ni, Z., Ji, X., Yi, J., Li, N., Wang, H., Liu, Y., & Yan, Y. (2026). Measurement-Efficient Few-Shot Vibration Fault Diagnosis via Physics-Informed Self-Supervised Learning and Adaptive Early Stopping. Sensors, 26(13), 4252. https://doi.org/10.3390/s26134252

Vancouver

Ni Z, Ji X, Yi J, Li N, Wang H, Liu Y, et al. Measurement-Efficient Few-Shot Vibration Fault Diagnosis via Physics-Informed Self-Supervised Learning and Adaptive Early Stopping. Sensors. 2026;26(13):4252. doi:10.3390/s26134252

BibTeX
@article{Ni2026,
  author = {Ni, Zongzhe and Ji, Xiancheng and Yi, Jianjun and Li, Nuozhou and Wang, Hongxing and Liu, Yifan and Yan, Ying},
  title = {Measurement-Efficient Few-Shot Vibration Fault Diagnosis via Physics-Informed Self-Supervised Learning and Adaptive Early Stopping},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {13},
  pages = {4252},
  doi = {10.3390/s26134252}
}
Kostenlos starten →