A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables

Xiaosheng Peng, Fan Yang, Ganjun Wang, Yijiang Wu, Lee Li, Zhaohui Li, Ashfaque Ahmed Bhatti, Chengke Zhou, Donald M. Hepburn, Alistair J. Reid, Martin D. Judd, Wah Hoon Siew

Research output: Contribution to journalArticle

9 Downloads (Pure)

Abstract

It is a great challenge to differentiate Partial Discharge (PD) induced by different types of insulation defect in high voltage cables. Some types of PD signals are very similar characteristics and are specifically difficult to be differentiate, even for the most experienced specialists. To overcome the challenge, a Convolutional Neural Network (CNN) based deep learning methodology for PD pattern recognition is presented in this paper. Firstly, PD testing for five types of artificial defects of ethylene-propylene-rubber (EPR) cables is carried out in the High Voltage (HV) laboratory to generate signals containing PD data. Secondly, 3500 sets of PD transient pulses are extracted and then 34 kinds of PD features are established. The third stage applies a CNN to the data: typical CNN architecture and the key factors which affect the CNN based pattern recognition accuracy, are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. The paper presents the flowchart of the CNN based PD pattern recognition method and the evaluation with 3500 sets of PD samples. Finally, the CNN based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e. Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.
Original languageEnglish
Pages (from-to)1460-1469
Number of pages10
JournalIEEE Transactions on Power Delivery
Volume34
Issue number4
Early online date19 Mar 2019
DOIs
Publication statusE-pub ahead of print - 19 Mar 2019

Fingerprint

Partial discharges
Cables
Neural networks
Electric potential
Pattern recognition
Backpropagation
Support vector machines
Deep learning
Defects
Network layers
Network architecture
Industrial applications
Propylene
Insulation
Rubber
Ethylene
Chemical activation

Keywords

  • convolutional neural network
  • deep learning
  • high voltage cables
  • partial discharge
  • pattern recognition

Cite this

Peng, Xiaosheng ; Yang, Fan ; Wang, Ganjun ; Wu, Yijiang ; Li, Lee ; Li, Zhaohui ; Bhatti, Ashfaque Ahmed ; Zhou, Chengke ; Hepburn, Donald M. ; Reid, Alistair J. ; Judd, Martin D. ; Siew, Wah Hoon. / A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables. In: IEEE Transactions on Power Delivery. 2019 ; Vol. 34, No. 4. pp. 1460-1469.
@article{709edd25337f4a7ba1955128b877cc95,
title = "A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables",
abstract = "It is a great challenge to differentiate Partial Discharge (PD) induced by different types of insulation defect in high voltage cables. Some types of PD signals are very similar characteristics and are specifically difficult to be differentiate, even for the most experienced specialists. To overcome the challenge, a Convolutional Neural Network (CNN) based deep learning methodology for PD pattern recognition is presented in this paper. Firstly, PD testing for five types of artificial defects of ethylene-propylene-rubber (EPR) cables is carried out in the High Voltage (HV) laboratory to generate signals containing PD data. Secondly, 3500 sets of PD transient pulses are extracted and then 34 kinds of PD features are established. The third stage applies a CNN to the data: typical CNN architecture and the key factors which affect the CNN based pattern recognition accuracy, are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. The paper presents the flowchart of the CNN based PD pattern recognition method and the evaluation with 3500 sets of PD samples. Finally, the CNN based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e. Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.",
keywords = "convolutional neural network, deep learning, high voltage cables, partial discharge, pattern recognition",
author = "Xiaosheng Peng and Fan Yang and Ganjun Wang and Yijiang Wu and Lee Li and Zhaohui Li and Bhatti, {Ashfaque Ahmed} and Chengke Zhou and Hepburn, {Donald M.} and Reid, {Alistair J.} and Judd, {Martin D.} and Siew, {Wah Hoon}",
note = "Author advised that paper was accepted in early March 2019; have therefore estimated acceptance date as 01/03/19. 26/03/19 DC.",
year = "2019",
month = "3",
day = "19",
doi = "10.1109/TPWRD.2019.2906086",
language = "English",
volume = "34",
pages = "1460--1469",
journal = "IEEE Transactions on Power Delivery",
issn = "0885-8977",
publisher = "IEEE",
number = "4",

}

Peng, X, Yang, F, Wang, G, Wu, Y, Li, L, Li, Z, Bhatti, AA, Zhou, C, Hepburn, DM, Reid, AJ, Judd, MD & Siew, WH 2019, 'A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables', IEEE Transactions on Power Delivery, vol. 34, no. 4, pp. 1460-1469. https://doi.org/10.1109/TPWRD.2019.2906086

A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables. / Peng, Xiaosheng; Yang, Fan; Wang, Ganjun; Wu, Yijiang; Li, Lee; Li, Zhaohui; Bhatti, Ashfaque Ahmed; Zhou, Chengke; Hepburn, Donald M.; Reid, Alistair J.; Judd, Martin D.; Siew, Wah Hoon.

In: IEEE Transactions on Power Delivery, Vol. 34, No. 4, 19.03.2019, p. 1460-1469.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A convolutional neural network based deep learning methodology for recognition of partial discharge patterns from high voltage cables

AU - Peng, Xiaosheng

AU - Yang, Fan

AU - Wang, Ganjun

AU - Wu, Yijiang

AU - Li, Lee

AU - Li, Zhaohui

AU - Bhatti, Ashfaque Ahmed

AU - Zhou, Chengke

AU - Hepburn, Donald M.

AU - Reid, Alistair J.

AU - Judd, Martin D.

AU - Siew, Wah Hoon

N1 - Author advised that paper was accepted in early March 2019; have therefore estimated acceptance date as 01/03/19. 26/03/19 DC.

PY - 2019/3/19

Y1 - 2019/3/19

N2 - It is a great challenge to differentiate Partial Discharge (PD) induced by different types of insulation defect in high voltage cables. Some types of PD signals are very similar characteristics and are specifically difficult to be differentiate, even for the most experienced specialists. To overcome the challenge, a Convolutional Neural Network (CNN) based deep learning methodology for PD pattern recognition is presented in this paper. Firstly, PD testing for five types of artificial defects of ethylene-propylene-rubber (EPR) cables is carried out in the High Voltage (HV) laboratory to generate signals containing PD data. Secondly, 3500 sets of PD transient pulses are extracted and then 34 kinds of PD features are established. The third stage applies a CNN to the data: typical CNN architecture and the key factors which affect the CNN based pattern recognition accuracy, are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. The paper presents the flowchart of the CNN based PD pattern recognition method and the evaluation with 3500 sets of PD samples. Finally, the CNN based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e. Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.

AB - It is a great challenge to differentiate Partial Discharge (PD) induced by different types of insulation defect in high voltage cables. Some types of PD signals are very similar characteristics and are specifically difficult to be differentiate, even for the most experienced specialists. To overcome the challenge, a Convolutional Neural Network (CNN) based deep learning methodology for PD pattern recognition is presented in this paper. Firstly, PD testing for five types of artificial defects of ethylene-propylene-rubber (EPR) cables is carried out in the High Voltage (HV) laboratory to generate signals containing PD data. Secondly, 3500 sets of PD transient pulses are extracted and then 34 kinds of PD features are established. The third stage applies a CNN to the data: typical CNN architecture and the key factors which affect the CNN based pattern recognition accuracy, are described. Factors discussed include the number of the network layers, convolutional kernel size, activation function, and pooling method. The paper presents the flowchart of the CNN based PD pattern recognition method and the evaluation with 3500 sets of PD samples. Finally, the CNN based pattern recognition results are shown and the proposed method is compared with two more traditional analysis methods, i.e. Support Vector Machine (SVM) and Back Propagation Neural Network (BPNN). The results show that the proposed CNN method has higher pattern recognition accuracy than SVM and BPNN, and that the novel method is especially effective for PD type recognition in cases of signals of high similarity, which is applicable for industrial applications.

KW - convolutional neural network

KW - deep learning

KW - high voltage cables

KW - partial discharge

KW - pattern recognition

U2 - 10.1109/TPWRD.2019.2906086

DO - 10.1109/TPWRD.2019.2906086

M3 - Article

VL - 34

SP - 1460

EP - 1469

JO - IEEE Transactions on Power Delivery

JF - IEEE Transactions on Power Delivery

SN - 0885-8977

IS - 4

ER -