A Fast Asymmetric Network Model for Anomaly Monitoring of Industrial Processes
IEEE Access · 2026 · 14 · 94689-94695
DOI: 10.1109/access.2026.3705491 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Pei Liu und 1 weiteren Autor:innen. Die Arbeit nennt 32 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Industrial processes provide the core driving force for intelligent productions. The measured data are nonlinear and non-Gaussian, and the sampling frequency of the sensor is high in actual working conditions. Directly using the multivariate statistical method is challenging to obtain the ideal anomaly detection (AD) result. Therefore, this paper proposes an asymmetric neural network based on broad learning. This asymmetric neural network can study the correlation among variables. The optimal AD capability of this asymmetric neural network is proved theoretically. A reliable threshold can be obtained quickly using the golden section method, making the proposed algorithm possess powerful AD capability. This threshold can maximize the anomaly detection rate of the algorithm while ensuring an acceptable false alarm rate. Compared with other neural network-based methods, this method can improve computational efficiency and ensure good anomaly detection ability. Finally, the availability of this asymmetric neural network is verified on the simulation platform of the industrial process control system.
Zitieren
APA
Liu, P., & Ma, D. (2026). A Fast Asymmetric Network Model for Anomaly Monitoring of Industrial Processes. IEEE Access, 14, 94689-94695. https://doi.org/10.1109/access.2026.3705491
Vancouver
Liu P, Ma D. A Fast Asymmetric Network Model for Anomaly Monitoring of Industrial Processes. IEEE Access. 2026;14:94689-94695. doi:10.1109/access.2026.3705491
BibTeX
@article{Liu2026,
author = {Liu, Pei and Ma, Donghui},
title = {A Fast Asymmetric Network Model for Anomaly Monitoring of Industrial Processes},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {94689--94695},
doi = {10.1109/access.2026.3705491}
}
EUVIMED