EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Evaluation of Partial Discharge Separation Techniques

João Victor Jales Da Silva Melo, George Rossany Soares De Lira, Edson Guedes Da Costa, Pablo Bezerra Vilar, Antonio Francisco Leite Neto, Igor Monteiro Abreu Dos Santos, Hebert Santos Crispim, Fernando Do Nascimento Madia

IEEE Access · 2026 · 14 · 102074-102091

DOI: 10.1109/access.2026.3709996 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von João Victor Jales Da Silva Melo und 7 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Coordination for the Improvement of Higher Education Personnel, National Council for Scientific and Technological Development, ANEEL R&D project.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Partial discharge (PD) analysis is widely recognized as a valuable technique for detecting faults and evaluating the condition of electrical system assets. However, its practical application poses significant challenges. In real-world scenarios, PD detection can be hindered by external noise and the simultaneous occurrence of multiple PD types, such as corona, surface, and internal discharges, making it difficult to isolate the origin of the signals. To address this, signal separation techniques capable of distinguishing between different PD sources are essential. Although several techniques have been proposed in the literature for this purpose, few studies provide a comprehensive comparative evaluation. As a result, selecting an appropriate technique for practical applications remains a challenge. Most works propose separation techniques based on limited datasets, without defining their application constraints or how well they perform in comparison to other existing techniques. In this context, the present study introduces two extensive, publicly available databases of controlled PD measurements encompassing various discharge types recorded both individually and simultaneously. It also conducts a systematic comparative analysis of the most prominent PD source separation techniques reported in the literature. The evaluation relies on statistical metrics that assess each method’s ability to form well-defined and distinct clusters. To support this analysis, dedicated experimental setups were designed to generate partial discharges in a laboratory environment, and measurements were carried out using high-frequency current transformers (HFCTs) for internal, surface, and corona discharges. The results indicate that frequency-related features consistently achieve the highest separation performance across all evaluated techniques. Building on this finding, a feature selection procedure identified an optimal subset of five frequency-related attributes that outperforms all individual techniques evaluated. The selected feature set was further validated on a simultaneous discharge measurement, where an HDBSCAN clustering model successfully separated corona and internal discharge sources without prior knowledge of the number of sources, demonstrating its applicability in realistic monitoring scenarios.

Förderung

  • Coordination for the Improvement of Higher Education Personnel
  • National Council for Scientific and Technological Development
  • ANEEL R&D project

Zitieren

APA

Melo, J. V. J. D. S., Lira, G. R. S. D., Costa, E. G. D., Vilar, P. B., Neto, A. F. L., Santos, I. M. A. D., Crispim, H. S., & Madia, F. D. N. (2026). Evaluation of Partial Discharge Separation Techniques. IEEE Access, 14, 102074-102091. https://doi.org/10.1109/access.2026.3709996

Vancouver

Melo JVJDS, Lira GRSD, Costa EGD, Vilar PB, Neto AFL, Santos IMAD, et al. Evaluation of Partial Discharge Separation Techniques. IEEE Access. 2026;14:102074-102091. doi:10.1109/access.2026.3709996

BibTeX
@article{Melo2026,
  author = {Melo, João Victor Jales Da Silva and Lira, George Rossany Soares De and Costa, Edson Guedes Da and Vilar, Pablo Bezerra and Neto, Antonio Francisco Leite and Santos, Igor Monteiro Abreu Dos and Crispim, Hebert Santos and Madia, Fernando Do Nascimento},
  title = {Evaluation of Partial Discharge Separation Techniques},
  journal = {IEEE Access},
  year = {2026},
  volume = {14},
  pages = {102074--102091},
  doi = {10.1109/access.2026.3709996}
}
Kostenlos starten →