EUVIMEDEUVIMED entdecken
Crossref

A Novel Interpretable Multilayer Voting Network for Fault Diagnosis

Ying Yan, Guanting Liu, Jun Cai, Edmond Qi Wu, Adrian David Cheok, Aiguo Song

IEEE Transactions on Reliability · 2026 · 75 · 2710-2720

DOI: 10.1109/tr.2026.3714267 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Reliability, 2026. Verfasst von Ying Yan und 5 weiteren Autor:innen. Die Arbeit nennt 40 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Jiangsu Province, Natural Science Foundation of Anhui Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep learning models are widely used for fault diagnosis due to their complex pattern-capturing ability, but their lack of interpretability restricts practical application in critical systems. There is a growing need for reliable and interpretable fault diagnosis methods. In this article, a novel interpretable multilayer voting network (IMLVN) is proposed, aiming to boost classification accuracy and interpretability of fault diagnosis. Inspired by the “Mafia” board game’s voting-to-identify-the-killer mechanism, IMLVN applies a similar voting approach in a network to locate faulty devices and enhance interpretability. It has eight layers, each with a distinct function and clear physical meaning. For instance, the polynomial layer approximates fault variables, and the voting value generation layer gets voting values via a new activation function. This design improves nonlinear characteristic-capturing, aggregates information through voting, and clarifies the relationship between fault features and diagnostic results, achieving interpretability. Validated with real HVAC system data under a stricter chronological blocked protocol, IMLVN maintains very strong classification accuracy while improving evaluation rigor. The voting mechanism also boosts decision transparency by showing device voting relationships to identify fault sources, increasing user trust and offering a reliable, intuitive fault diagnosis solution. The research provides a new fault diagnosis perspective for complex networks and has significant theoretical and practical value.

Förderung

  • National Natural Science Foundation of China
  • Natural Science Foundation of Jiangsu Province
  • Natural Science Foundation of Anhui Province
  • Open Foundation of the Key Laboratory of Technology and Equipment of Tianjin
  • Excellent Research and Innovation Team Project of Universities in Anhui Province
  • Research on Quality Assurance and Evaluation of Higher Education in Jiangsu Province

Zitieren

APA

Yan, Y., Liu, G., Cai, J., Wu, E. Q., Cheok, A. D., & Song, A. (2026). A Novel Interpretable Multilayer Voting Network for Fault Diagnosis. IEEE Transactions on Reliability, 75, 2710-2720. https://doi.org/10.1109/tr.2026.3714267

Vancouver

Yan Y, Liu G, Cai J, Wu EQ, Cheok AD, Song A. A Novel Interpretable Multilayer Voting Network for Fault Diagnosis. IEEE Transactions on Reliability. 2026;75:2710-2720. doi:10.1109/tr.2026.3714267

BibTeX
@article{Yan2026,
  author = {Yan, Ying and Liu, Guanting and Cai, Jun and Wu, Edmond Qi and Cheok, Adrian David and Song, Aiguo},
  title = {A Novel Interpretable Multilayer Voting Network for Fault Diagnosis},
  journal = {IEEE Transactions on Reliability},
  year = {2026},
  volume = {75},
  pages = {2710--2720},
  doi = {10.1109/tr.2026.3714267}
}
Kostenlos starten →