Classification of Common Discharges in Outdoor Insulators Using Ultrasonic Signals

Ceramic insulators have been in use for more than a century, and a significant portion of them have approached or exceeded their intended lifetime. Consequently, health monitoring of outdoor insulation systems is crucial to the integrity of the transmission and distribution overhead lines. This stud...

Full description

Saved in:
Bibliographic Details
Main Authors: Lutfi, Abdulla Eyad, El-Hag, Ayman, Shaban, Khaled
Format: Conference Proceeding
Language:English
Subjects:
Acoustics
Ceramics
Convolutional Neural Networks
Discharges (electric)
Dry Band Arcing
Insulators
Neural networks
Partial Discharge
Sensors
Ultrasonic Sensor
Ultrasonic variables measurement
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Ceramic insulators have been in use for more than a century, and a significant portion of them have approached or exceeded their intended lifetime. Consequently, health monitoring of outdoor insulation systems is crucial to the integrity of the transmission and distribution overhead lines. This study uses nonintrusive techniques based on ultrasonic signal measurements and deep convolutional neural networks (CNN) to classify different discharges in defective outdoor insulators. Four different defects, namely, corona, internal void, external dry, and wet discharges, are introduced to insulator samples under controlled laboratory conditions. The CNN model is trained on these discharges and tested using unseen signals to classify the type of defect. The model achieved an overall accuracy of 97.95% showing superiority over traditional machine learning models.
ISSN:2576-2397
DOI:10.1109/CEIDP55452.2022.9985259