EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Pattern Recognition of GIS Partial Discharge Based on UHF Signal Characteristics

Shaoming Pan, Wei Zhang, Yuan Ma, Yi Su, Wei Huang

Electronics · 06.03.2026 · 15 (5) · 1096

DOI: 10.3390/electronics15051096 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Electronics, online erschienen am 06.03.2026. Verfasst von Shaoming Pan und 4 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key Research and Development Program of China, Science and Technology Project of Guangxi Electric Power Grid Corporation.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The partial discharge (PD) caused by insulation defects of gas-insulated switchgear (GIS) threatens the secure and stable operation of power systems. Traditional PD pattern recognition methods exhibit limitations due to incomplete information utilization and unresolved correlations among characteristic parameters. Based on the partial discharge mechanisms of GIS, this paper establishes a GIS partial discharge simulation model using the finite element time-domain (FETD) method. The propagation rules and influence factors of ultra-high-frequency (UHF) signals are studied. Furthermore, a PD pattern recognition method based on a deep convolutional neural network (CNN) is proposed. Research results indicate that UHF signals generated by GIS partial discharge are significantly influenced by pulse current waveforms and discharge quantity. The peak-to-peak amplitude of the electric field (Epp) increases linearly with the current amplitude, while it decreases nonlinearly with increasing pulse width. The UHF signal remains a certain value while the pulse width exceeds a critical threshold (4 ns). The proposed CNN-based approach, utilizing full-wave UHF signals, overcomes the shortcomings of traditional methods reliant on manually extracted discrete feature parameters. Compared to other network architectures and optimization algorithms, the ConvNeXt-AdamW model demonstrates superior performance, achieving an average PD pattern recognition accuracy exceeding 96%.

Förderung

  • National Key Research and Development Program of China
  • Science and Technology Project of Guangxi Electric Power Grid Corporation

Zitieren

APA

Pan, S., Zhang, W., Ma, Y., Su, Y., & Huang, W. (2026). Pattern Recognition of GIS Partial Discharge Based on UHF Signal Characteristics. Electronics, 15(5), 1096. https://doi.org/10.3390/electronics15051096

Vancouver

Pan S, Zhang W, Ma Y, Su Y, Huang W. Pattern Recognition of GIS Partial Discharge Based on UHF Signal Characteristics. Electronics. 2026;15(5):1096. doi:10.3390/electronics15051096

BibTeX
@article{Pan2026,
  author = {Pan, Shaoming and Zhang, Wei and Ma, Yuan and Su, Yi and Huang, Wei},
  title = {Pattern Recognition of GIS Partial Discharge Based on UHF Signal Characteristics},
  journal = {Electronics},
  year = {2026},
  volume = {15},
  number = {5},
  pages = {1096},
  doi = {10.3390/electronics15051096}
}
Kostenlos starten →