Discharges Classification using Genetic Algorithms and Feature Selection Algorithms on Time and Frequency Domain Data Extracted from Leakage Current Measurements

Discharges Classification using Genetic Algorithms and Feature Selection Algorithms on Time and Frequency Domain Data Extracted from Leakage Current Measurements

Show other versions (1)

A number of 387 discharge portraying waveforms recorded on 18 different 150 kV post insulators installed at two different Substations in Crete, Greece are considered in this paper. Twenty different features are extracted from each waveform and two feature selection algorithms (t-test and mRMR) are e...

Full description

Bibliographic Details
Main Authors: D. Pylarinos, K. Theofilatos, K. Siderakis, E. Thalassinakis
Format: Article
Language:English
Published: D. G. Pylarinos 2013-12-01
Series:Engineering, Technology & Applied Science Research
Subjects:
insulators
leakage current
discharges
classification
frequency
genetic algorithms
Online Access:https://etasr.com/index.php/ETASR/article/view/418
Description
Summary:A number of 387 discharge portraying waveforms recorded on 18 different 150 kV post insulators installed at two different Substations in Crete, Greece are considered in this paper. Twenty different features are extracted from each waveform and two feature selection algorithms (t-test and mRMR) are employed. Genetic algorithms are used to classify waveforms in two different classes related to the portrayed discharges. Five different data sets are employed (1. the original feature vector, 2. time domain features, 3. frequency domain features, 4. t-test selected features 5. mRMR selected features). Results are discussed and compared with previous classification implementations on this particular data group.
ISSN:2241-4487
1792-8036

Similar Items