Neural network based diagnosis of partial discharge defects patterns at XLPE cable under DC stress

The application of the XLPE cable under DC stress is now being considered as one of the promising tool for enlarging the limited transmission capacity of the traditional electric power cable due to the several technical advantages. Therefore, many research institutes are trying to improve the perfor...

Full description

Saved in:
Bibliographic Details
Published in:Electrical engineering 2017-03, Vol.99 (1), p.119-132
Main Authors: Khan, Muhammad Yousaf Ali, Koo, Ja-Yoon
Format: Article
Language:English
Subjects:
Defects
Economics and Management
Electric power transmission
Electrical Engineering
Electrical Machines and Networks
Energy Policy
Engineering
Feature extraction
Frequency domain analysis
Insulation
Neural networks
Original Paper
Pattern recognition
Performance enhancement
Power cables
Power Electronics
Research facilities
Stresses
Training
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The application of the XLPE cable under DC stress is now being considered as one of the promising tool for enlarging the limited transmission capacity of the traditional electric power cable due to the several technical advantages. Therefore, many research institutes are trying to improve the performance by carrying out experimental investigations related to the reliability of the cable to be used under DC stress. One of them is partial discharge (PD) detection, which is considered as the tool to indicate the insulation state of the apparatus; however, very few reports have been published based on the results obtained for the XLPE cable used under the DC stress. In this work, three different types of artificial defects are designed to produce PD under DC: void, surface and corona discharge defects. Pattern recognition of the detected PD signal is based on modified CAPD method. To improve the PD pattern recognition rate, the training data obtained through neural networks is combined with the results of the frequency domain analysis. Finally, the training data and mean square error of PD patterns obtained through neural networks is compared with the training data obtained with the use of only modified CAPD method.
ISSN:0948-7921
1432-0487
DOI:10.1007/s00202-016-0395-0