Robust Classification of PD Sources Using Deep Learning and Signal Processing Techniques

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F24%3A00378780" target="_blank" >RIV/68407700:21230/24:00378780 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1109/Diagnostika61830.2024.10693892" target="_blank" >https://doi.org/10.1109/Diagnostika61830.2024.10693892</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/Diagnostika61830.2024.10693892" target="_blank" >10.1109/Diagnostika61830.2024.10693892</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Robust Classification of PD Sources Using Deep Learning and Signal Processing Techniques

Popis výsledku v původním jazyce

Partial discharge (PD) pattern recognition technologies are highly valuable for the predictive maintenance and diagnosis of electrical insulation systems. PD signals are generated when the voltage across the cavity embedded in the insulation system exceeds the PD inception voltage. These signals are low-energy, high-frequency, and stochastic, making analysis extremely challenging since common signal processing methods cannot analyze and interpret their transient, non-stationary characteristics. It is crucial to address PD pattern recognition with a comprehensive strategy that includes collecting PD databases, feature extraction, and applying proper machine learning models. Using cutting-edge PD signal processing techniques and deep learning, PD patterns could be successfully analyzed and interpreted, resulting in the early identification of PD sources inside insulation and reducing the likelihood of system failure. The aim of the paper is to investigate and evaluate various methods to recognize partial discharge (PD) signals as single PD source (SPS) or double PD sources (DPS). Wavelet Analysis, Short-Time Fourier Transform (STFT), Wigner-Ville Distribution (WVD), Empirical Mode Decomposition (EMD), and Higher-Order Statistics (HOS) are among the techniques used, along with deep learning models.

Název v anglickém jazyce

Robust Classification of PD Sources Using Deep Learning and Signal Processing Techniques

Popis výsledku anglicky

Partial discharge (PD) pattern recognition technologies are highly valuable for the predictive maintenance and diagnosis of electrical insulation systems. PD signals are generated when the voltage across the cavity embedded in the insulation system exceeds the PD inception voltage. These signals are low-energy, high-frequency, and stochastic, making analysis extremely challenging since common signal processing methods cannot analyze and interpret their transient, non-stationary characteristics. It is crucial to address PD pattern recognition with a comprehensive strategy that includes collecting PD databases, feature extraction, and applying proper machine learning models. Using cutting-edge PD signal processing techniques and deep learning, PD patterns could be successfully analyzed and interpreted, resulting in the early identification of PD sources inside insulation and reducing the likelihood of system failure. The aim of the paper is to investigate and evaluate various methods to recognize partial discharge (PD) signals as single PD source (SPS) or double PD sources (DPS). Wavelet Analysis, Short-Time Fourier Transform (STFT), Wigner-Ville Distribution (WVD), Empirical Mode Decomposition (EMD), and Higher-Order Statistics (HOS) are among the techniques used, along with deep learning models.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20201 - Electrical and electronic engineering

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2024

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

2024 International Conference on Diagnostics in Electrical Engineering (Diagnostika)

ISBN

979-8-3503-6149-0

ISSN

2464-708X

e-ISSN

2464-708X

Počet stran výsledku

5

Strana od-do

117-121

Název nakladatele

Institute of Electrical and Electronics Engineers, Inc.

Místo vydání

—

Místo konání akce

Plzeň

Datum konání akce

3. 8. 2024

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

001345150300024