Zero-Shot Intelligent Fault Detection of Rotating Machinery Based on Physics-Embedded Constrained Mamba Network
IEEE Sensors Journal · 2026-08 · 26 (15) · 23190-23201
DOI: 10.1109/jsen.2026.3706798 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Sensors Journal, 2026-08. Verfasst von Jiajing Xi und 5 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Gefördert von: Natural Science Foundation of Fujian Province, Natural Science Foundation of Xiamen, Key Laboratory of Marine Power Engineering Technology of Ministry of Transport.
Abstract (Originalsprache)
High costs of reliability testing, insufficient real-world fault samples, and distribution shifts in laboratory data impede the development of efficient data-driven condition monitoring models for rotating machinery. To address the above challenges, this article proposed a zero-shot intelligent anomaly detection method for rotating machinery driven by the time–frequency prior knowledge and a Mamba network designed with a new self-attention module. First, multiple time–frequency analysis methods are introduced to deal with the normal condition real-time monitoring signals of rotating machinery. Then, the strong temporal characteristics of the monitored signals and the inherent physical differences among diverse time–frequency representations similarity and dissimilarity are mathematically redefined, which are used to develop three novel contrastive learning loss functions without any fault samples. Subsequently, a novel self-attention module is designed to establish the Mamba network, which is applied as the high-dimensional feature encoder for the diverse time–frequency representations. Finally, a physical knowledge-based anomaly detection algorithm is established to capture the fault condition of rotating machinery. Two rotating machinery test platforms are used to demonstrate the effectiveness of the proposed method, which achieves high-precision anomaly detection without fault samples across diverse rotating machinery.
Förderung
- Natural Science Foundation of Fujian Province
- Natural Science Foundation of Xiamen
- Key Laboratory of Marine Power Engineering Technology of Ministry of Transport
Zitieren
APA
Xi, J., Li, G., Xu, L., Cheng, Y., Wu, D., & Shao, H. (2026). Zero-Shot Intelligent Fault Detection of Rotating Machinery Based on Physics-Embedded Constrained Mamba Network. IEEE Sensors Journal, 26(15), 23190-23201. https://doi.org/10.1109/jsen.2026.3706798
Vancouver
Xi J, Li G, Xu L, Cheng Y, Wu D, Shao H. Zero-Shot Intelligent Fault Detection of Rotating Machinery Based on Physics-Embedded Constrained Mamba Network. IEEE Sensors Journal. 2026;26(15):23190-23201. doi:10.1109/jsen.2026.3706798
BibTeX
@article{Xi2026,
author = {Xi, Jiajing and Li, Guoqiang and Xu, Lijie and Cheng, Yiwei and Wu, Defeng and Shao, Haidong},
title = {Zero-Shot Intelligent Fault Detection of Rotating Machinery Based on Physics-Embedded Constrained Mamba Network},
journal = {IEEE Sensors Journal},
year = {2026},
volume = {26},
number = {15},
pages = {23190--23201},
doi = {10.1109/jsen.2026.3706798}
}
EUVIMED