Identification of Partial Discharges at DC Voltage using Machine Learning Methods

Publikation: KonferenzbeitragPaperForschungBegutachtung

Abstract

In order to guarantee the operational reliability of high voltage direct current
equipment, the partial discharge measurement plays an important role. For AC systems,
numerous works on partial discharge interpretation have been carried out for several
decades. However, these methods for AC systems cannot be applied directly to DC
systems. In this work, partial discharge data for typical defects in gas-insulated systems
were accumulated and several identification methods on partial discharges at DC voltage
were compared. As for the experiment, specially designed test cells were used, including
three kinds of typical defects: floating electrode, protrusion on a high voltage electrode and
free metallic particle. The experiments were performed in SF6 gas at DC voltages with both
positive and negative polarity. The experimental results showed that even in the same
category of the defect, such as the floating electrode, the partial discharge patterns greatly
varied depending on test conditions or the shape of the defect. As for the partial discharge
identification, three kinds of input data (statistical features, raw partial discharge data and
the pixel data of the NoDi* Pattern mappings) and two kinds of identification algorithms (the
artificial neural network and the decision tree) were combined. The performances of all
these methods were compared. Using the statistical features or the pixel data of NoDi*
Pattern mappings as the input data showed good performances and were able to correctly
identify PD defects with more than 95 %.
Originalspracheenglisch
Seitenumfang6
PublikationsstatusVeröffentlicht - 1 Sep 2017
Veranstaltung20th International Symposium on High Voltage Engineering - Buenos Aires, Argentinien
Dauer: 27 Aug 20171 Sep 2017
http://www.ish2017.org/

Konferenz

Konferenz20th International Symposium on High Voltage Engineering
KurztitelISH 2017
LandArgentinien
OrtBuenos Aires
Zeitraum27/08/171/09/17
Internetadresse

Fields of Expertise

  • Sustainable Systems

Dies zitieren

Kainaga, Pirker, A., & Schichler, U. (2017). Identification of Partial Discharges at DC Voltage using Machine Learning Methods. Beitrag in 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentinien.

Identification of Partial Discharges at DC Voltage using Machine Learning Methods. / Kainaga, ; Pirker, Alexander; Schichler, Uwe.

2017. Beitrag in 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentinien.

Publikation: KonferenzbeitragPaperForschungBegutachtung

Kainaga, , Pirker, A & Schichler, U 2017, 'Identification of Partial Discharges at DC Voltage using Machine Learning Methods' Beitrag in 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentinien, 27/08/17 - 1/09/17, .
Kainaga , Pirker A, Schichler U. Identification of Partial Discharges at DC Voltage using Machine Learning Methods. 2017. Beitrag in 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentinien.
Kainaga, ; Pirker, Alexander ; Schichler, Uwe. / Identification of Partial Discharges at DC Voltage using Machine Learning Methods. Beitrag in 20th International Symposium on High Voltage Engineering, Buenos Aires, Argentinien.6 S.
@conference{b1512b8ae1ed4041ad2d35c18005b7d2,
title = "Identification of Partial Discharges at DC Voltage using Machine Learning Methods",
abstract = "In order to guarantee the operational reliability of high voltage direct currentequipment, the partial discharge measurement plays an important role. For AC systems,numerous works on partial discharge interpretation have been carried out for severaldecades. However, these methods for AC systems cannot be applied directly to DCsystems. In this work, partial discharge data for typical defects in gas-insulated systemswere accumulated and several identification methods on partial discharges at DC voltagewere compared. As for the experiment, specially designed test cells were used, includingthree kinds of typical defects: floating electrode, protrusion on a high voltage electrode andfree metallic particle. The experiments were performed in SF6 gas at DC voltages with bothpositive and negative polarity. The experimental results showed that even in the samecategory of the defect, such as the floating electrode, the partial discharge patterns greatlyvaried depending on test conditions or the shape of the defect. As for the partial dischargeidentification, three kinds of input data (statistical features, raw partial discharge data andthe pixel data of the NoDi* Pattern mappings) and two kinds of identification algorithms (theartificial neural network and the decision tree) were combined. The performances of allthese methods were compared. Using the statistical features or the pixel data of NoDi*Pattern mappings as the input data showed good performances and were able to correctlyidentify PD defects with more than 95 {\%}.",
author = "Kainaga and Alexander Pirker and Uwe Schichler",
year = "2017",
month = "9",
day = "1",
language = "English",
note = "20th International Symposium on High Voltage Engineering, ISH 2017 ; Conference date: 27-08-2017 Through 01-09-2017",
url = "http://www.ish2017.org/",

}

TY - CONF

T1 - Identification of Partial Discharges at DC Voltage using Machine Learning Methods

AU - Kainaga,

AU - Pirker, Alexander

AU - Schichler, Uwe

PY - 2017/9/1

Y1 - 2017/9/1

N2 - In order to guarantee the operational reliability of high voltage direct currentequipment, the partial discharge measurement plays an important role. For AC systems,numerous works on partial discharge interpretation have been carried out for severaldecades. However, these methods for AC systems cannot be applied directly to DCsystems. In this work, partial discharge data for typical defects in gas-insulated systemswere accumulated and several identification methods on partial discharges at DC voltagewere compared. As for the experiment, specially designed test cells were used, includingthree kinds of typical defects: floating electrode, protrusion on a high voltage electrode andfree metallic particle. The experiments were performed in SF6 gas at DC voltages with bothpositive and negative polarity. The experimental results showed that even in the samecategory of the defect, such as the floating electrode, the partial discharge patterns greatlyvaried depending on test conditions or the shape of the defect. As for the partial dischargeidentification, three kinds of input data (statistical features, raw partial discharge data andthe pixel data of the NoDi* Pattern mappings) and two kinds of identification algorithms (theartificial neural network and the decision tree) were combined. The performances of allthese methods were compared. Using the statistical features or the pixel data of NoDi*Pattern mappings as the input data showed good performances and were able to correctlyidentify PD defects with more than 95 %.

AB - In order to guarantee the operational reliability of high voltage direct currentequipment, the partial discharge measurement plays an important role. For AC systems,numerous works on partial discharge interpretation have been carried out for severaldecades. However, these methods for AC systems cannot be applied directly to DCsystems. In this work, partial discharge data for typical defects in gas-insulated systemswere accumulated and several identification methods on partial discharges at DC voltagewere compared. As for the experiment, specially designed test cells were used, includingthree kinds of typical defects: floating electrode, protrusion on a high voltage electrode andfree metallic particle. The experiments were performed in SF6 gas at DC voltages with bothpositive and negative polarity. The experimental results showed that even in the samecategory of the defect, such as the floating electrode, the partial discharge patterns greatlyvaried depending on test conditions or the shape of the defect. As for the partial dischargeidentification, three kinds of input data (statistical features, raw partial discharge data andthe pixel data of the NoDi* Pattern mappings) and two kinds of identification algorithms (theartificial neural network and the decision tree) were combined. The performances of allthese methods were compared. Using the statistical features or the pixel data of NoDi*Pattern mappings as the input data showed good performances and were able to correctlyidentify PD defects with more than 95 %.

M3 - Paper

ER -