Multimodality Uncertainty-Aware Fault Diagnosis Framework for Small-Sample Bearing Faults Under Variable Operating Conditions With Strong Noises
IEEE Sensors Journal · 2026-08 · 26 (15) · 22760-22773
DOI: 10.1109/jsen.2026.3702987 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-08. Verfasst von Jiayu Liu und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Hunan Province, Special Program for Promoting Intellectual Property Strategy of Hunan Province.
Abstract (Originalsprache)
Fault diagnosis of bearings under varying working conditions and strong noise interference remains a significant challenge, especially when only small-sample data is available. To address these issues, this article proposes a multimodality uncertainty-aware fault diagnosis framework (MUFDF). Instead of relying on single-domain analysis or stacking redundant network components, MUFDF adaptively extracts and fuses temporal, frequency, and image-domain features using an attention mechanism, and employs a Bayesian neural network (BNN) to quantify prediction confidence. Experiments on two public bearing datasets [Case Western Reserve University (CWRU) and spectraquest (SQ)] and a real-world axial fan dataset validate the superiority of the proposed method. Specifically, under the extreme condition of only one training sample per class and a strong noise level of 0 dB signal-to-noise ratio (SNR), MUFDF achieved average classification accuracies of 97.18%, 94.62%, and 95.83% across the three datasets, respectively, significantly outperforming several state-of-the-art baseline models. The quantitative results confirm the method’s robustness and practical applicability in complex industrial scenarios.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Hunan Province
- Special Program for Promoting Intellectual Property Strategy of Hunan Province
- Scientific Research Project of Hunan Provincial Department of Education
Zitieren
APA
Liu, J., Liu, J., Xiao, J., Li, J., & Shao, H. (2026). Multimodality Uncertainty-Aware Fault Diagnosis Framework for Small-Sample Bearing Faults Under Variable Operating Conditions With Strong Noises. IEEE Sensors Journal, 26(15), 22760-22773. https://doi.org/10.1109/jsen.2026.3702987
Vancouver
Liu J, Liu J, Xiao J, Li J, Shao H. Multimodality Uncertainty-Aware Fault Diagnosis Framework for Small-Sample Bearing Faults Under Variable Operating Conditions With Strong Noises. IEEE Sensors Journal. 2026;26(15):22760-22773. doi:10.1109/jsen.2026.3702987
BibTeX
@article{Liu2026,
author = {Liu, Jiayu and Liu, Jinping and Xiao, Jingri and Li, Jinyan and Shao, Haidong},
title = {Multimodality Uncertainty-Aware Fault Diagnosis Framework for Small-Sample Bearing Faults Under Variable Operating Conditions With Strong Noises},
journal = {IEEE Sensors Journal},
year = {2026},
volume = {26},
number = {15},
pages = {22760--22773},
doi = {10.1109/jsen.2026.3702987}
}
EUVIMED