EUVIMEDEUVIMED entdecken
Crossref

Acoustic fault diagnosis in heterogeneous environments via causal disentanglement

Ruichen Zhai, Zhongliang Wei, Chang Su

Measurement Science and Technology · 26.06.2026 · 37 (26) · 266103

DOI: 10.1088/1361-6501/ae7aaa ↗

Kurz gefasst

Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 25.06.2026. Verfasst von Ruichen Zhai und 2 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Gefördert von: National Natural Science Foundation of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Non-contact acoustic sensing offers flexible monitoring but suffers from systematic uncertainties caused by heterogeneous configurations and complex transmission channels. Conventional methods often conflate intrinsic fault signatures with environmental artifacts, leading to sensitivity to measurement-induced variations. To address this calibration challenge, we propose the Causal Bilinear Generalization Network (CBGN), a physics-grounded framework anchored in structural causal models. Unlike methods relying on superficial statistical alignment, CBGN explicitly models the acquired signal as a coupled outcome of invariant source excitations and configuration-dependent channel responses. We devise a two-stage optimization strategy to perform blind identification of the fault mechanism, employing compact bilinear pooling to mathematically capture the non-linear modulation inherent to acoustic propagation. This ensures diagnostic decision boundaries remain invariant to measurement setup variations. Extensive experiments demonstrate that CBGN achieves state-of-the-art performance, exhibiting superior measurement stability particularly under severe spectral interference and path loss. The source code is released at https://github.com/Malones314/Causal-Bilinear-Generalization-Network .

Förderung

  • National Natural Science Foundation of China

Zitieren

APA

Zhai, R., Wei, Z., & Su, C. (2026). Acoustic fault diagnosis in heterogeneous environments via causal disentanglement. Measurement Science and Technology, 37(26), 266103. https://doi.org/10.1088/1361-6501/ae7aaa

Vancouver

Zhai R, Wei Z, Su C. Acoustic fault diagnosis in heterogeneous environments via causal disentanglement. Measurement Science and Technology. 2026;37(26):266103. doi:10.1088/1361-6501/ae7aaa

BibTeX
@article{Zhai2026,
  author = {Zhai, Ruichen and Wei, Zhongliang and Su, Chang},
  title = {Acoustic fault diagnosis in heterogeneous environments via causal disentanglement},
  journal = {Measurement Science and Technology},
  year = {2026},
  volume = {37},
  number = {26},
  pages = {266103},
  doi = {10.1088/1361-6501/ae7aaa}
}
Kostenlos starten →