EUVIMEDEUVIMED entdecken
Crossref

Automatic Power System Transient Instability Mode Identification via One-Class Deep Learning and Control Effect Validation

Jie Hou, Lipeng Zhu, Quan Zhou, Jiayong Li, Cong Zhang, Yunhe Hou

IEEE Transactions on Automation Science and Engineering · 2026 · 23 · 12402-12415

DOI: 10.1109/tase.2026.3703798 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Automation Science and Engineering, 2026. Verfasst von Jie Hou und 5 weiteren Autor:innen. Die Arbeit nennt 45 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Science and Technology Innovation Program of Hunan Province.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In practical power systems, transient rotor angle instability and short-term voltage instability are often strongly coupled, making it difficult to accurately identify transient dominant instability modes (TDIMs). To address this challenge, this paper proposes an automatic TDIM identification framework that integrates deep learning (DL) with control effect validation to achieve both high efficiency and reliability. First, transient cases with weak mode couplings are directly identified using conventional, widely adopted stability criteria. For strongly coupled cases, TDIMs are ascertained through control effect validation that evaluates the stabilization effectiveness of candidate control actions. To further improve computational efficiency, a one-class DL strategy is developed to predict stabilization results using early-stage system transient trajectories, enabling early termination of costly time-domain simulations. The efficacy of the proposed framework is comprehensively tested on the China EPRI 36-bus system and the IEEE 162-bus system. Results show that the framework can automatically identify TDIMs across diverse transient scenarios without compromising identification reliability. Besides, the one-class DL strategy can reduce the time required for the control effect validation process in the proposed TDIM identification framework by more than 30%. Note to Practitioners—This paper is motivated by the challenging problem of automatically identifying transient dominant instability modes (TDIMs) with both high reliability and efficiency for practical power systems. Following a large disturbance, practical power grids operated near stability limits often exhibit strong couplings between rotor angle (RAI) and short-term voltage instability (SVI). Such couplings make it difficult to determine system TDIMs clearly. As a result, it becomes challenging to decide which emergency control action should be taken, such as generator tripping against RAI or load shedding against SVI. Conventional manual inspection or model-based approaches are computationally expensive and rely heavily on expert experience, limiting their practicality for the screening of large numbers of contingencies. In this paper, an automatic TDIM identification approach is developed by combining one-class deep learning with control effect validation. With the combination of these two key components, TDIMs under diverse transient scenarios can be efficiently identified without losing reliability. The proposed approach is validated through comprehensive simulations on the China EPRI 36-bus benchmark system and is further examined on the IEEE 162-bus system, showing its potential applicability to larger-scale benchmark power systems. However, this work has not yet provided interpretable insights into the underlying dominant instability mechanisms. Future work will focus on developing an interpretable TDIM assessment scheme, which may help uncover the root causes of typical TDIMs.

Förderung

  • National Natural Science Foundation of China
  • Science and Technology Innovation Program of Hunan Province

Zitieren

APA

Hou, J., Zhu, L., Zhou, Q., Li, J., Zhang, C., & Hou, Y. (2026). Automatic Power System Transient Instability Mode Identification via One-Class Deep Learning and Control Effect Validation. IEEE Transactions on Automation Science and Engineering, 23, 12402-12415. https://doi.org/10.1109/tase.2026.3703798

Vancouver

Hou J, Zhu L, Zhou Q, Li J, Zhang C, Hou Y. Automatic Power System Transient Instability Mode Identification via One-Class Deep Learning and Control Effect Validation. IEEE Transactions on Automation Science and Engineering. 2026;23:12402-12415. doi:10.1109/tase.2026.3703798

BibTeX
@article{Hou2026,
  author = {Hou, Jie and Zhu, Lipeng and Zhou, Quan and Li, Jiayong and Zhang, Cong and Hou, Yunhe},
  title = {Automatic Power System Transient Instability Mode Identification via One-Class Deep Learning and Control Effect Validation},
  journal = {IEEE Transactions on Automation Science and Engineering},
  year = {2026},
  volume = {23},
  pages = {12402--12415},
  doi = {10.1109/tase.2026.3703798}
}
Kostenlos starten →