EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Bayesian Conditional GAN for Unsupervised Anomaly Detection in Structural Health Monitoring Time-Series Dataset

Yohannes L Alemu iD, Christian Walther iD, Manuel Schneider iD, Norbert Greifzu, Leon Quinten Thiebes, Andreas Wenzel, Uwe Plank-Wiedenbeck, Tom Lahmer iD

Sensors · 04.07.2026 · 26 (13) · 4253

DOI: 10.3390/s26134253 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 04.07.2026. Verfasst von Yohannes L Alemu und 7 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Structural Mechanics, Bauhaus University, 99423 Weimar, Germany. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Federal Ministry of Education and Research (BMBF), German Federal Ministry of Education and Research.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Detecting rare structural damage without labeled fault data remains a critical unsolved challenge in structural health monitoring (SHM). Prestressed concrete catenary poles are key elements of high-speed railway infrastructure, and undetected degradation can compromise safety and service reliability. This paper introduces BcDCGAN, a Bayesian conditional deep convolutional generative adversarial network designed for unsupervised anomaly detection in multivariate vibration time series from three in-service catenary poles. Trained exclusively on healthy acceleration signals with wind-speed conditioning, the model learns the normal structural dynamics and produces an uncertainty-based anomaly score that combines reconstruction quality, adversarial evaluation, and epistemic uncertainty into a single decision function. An adaptive, data-driven threshold estimate from healthy validation data enables practical deployment without damage labels. On a real 2017 catenary pole dataset (1606 signals, 70/10/20 split) with injected, physically motivated damage-like patterns, BcDCGAN achieves high anomaly recall with interpretable uncertainty signals and clear separation between normal and anomalous latent representations. The results suggest that Bayesian conditional GANs can support risk-aware monitoring of railway infrastructure under varying environmental and operational conditions.

Schlagwörter

Stichwörter der Autor:innen

Bayesian conditional deep convolutional generative adversarial networkBayesian inferenceconditional GANstructural health monitoringtemporal causal networksuncertainty quantificationunsupervised anomaly detection

Autor:innen und Institutionen

  • Institute of Structural Mechanics, Bauhaus University, 99423 Weimar, Germany.
  • Faculty of Electrical Engineering, Schmalkalden University of Applied Sciences, 98574 Schmalkalden, Germany.
  • Chair of Transport System Planning, Bauhaus-University Weimar, 99423 Weimar, Germany.

ORCID-Profile: Yohannes L Alemu, Christian Walther, Manuel Schneider, Tom Lahmer

Förderung

  • Federal Ministry of Education and Research (BMBF)
  • German Federal Ministry of Education and Research

Zitieren

APA

Alemu, Y. L., Walther, C., Schneider, M., Greifzu, N., Thiebes, L. Q., Wenzel, A., Plank-Wiedenbeck, U., & Lahmer, T. (2026). Bayesian Conditional GAN for Unsupervised Anomaly Detection in Structural Health Monitoring Time-Series Dataset. Sensors, 26(13), 4253. https://doi.org/10.3390/s26134253

Vancouver

Alemu YL, Walther C, Schneider M, Greifzu N, Thiebes LQ, Wenzel A, et al. Bayesian Conditional GAN for Unsupervised Anomaly Detection in Structural Health Monitoring Time-Series Dataset. Sensors. 2026;26(13):4253. doi:10.3390/s26134253

BibTeX
@article{Alemu2026,
  author = {Alemu, Yohannes L and Walther, Christian and Schneider, Manuel and Greifzu, Norbert and Thiebes, Leon Quinten and Wenzel, Andreas and Plank-Wiedenbeck, Uwe and Lahmer, Tom},
  title = {Bayesian Conditional GAN for Unsupervised Anomaly Detection in Structural Health Monitoring Time-Series Dataset},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {13},
  pages = {4253},
  doi = {10.3390/s26134253}
}
Kostenlos starten →