https://doi.org/10.1051/epjap/2012120050
ANN and wavelet-based discrimination technique between discharge currents in transformer mineral oils
1
Laboratory of Electrical and Industrial Systems, FEI, USTHB, BP 32 Bab Ezzouar, Algiers 16311, Algeria
2
École Centrale de Lyon, AMPERE Laboratory, UMR CNRS 5005, 36 avenue Guy de Collongue, 69134 Écully, France
a e-mail: Abderrahmane.Beroual@eea.ec-lyon.fr
Received:
8
February
2012
Revised:
29
March
2012
Accepted:
4
April
2012
Published online:
9
May
2012
This paper is aimed at the analysis of positive pre-breakdown currents triggered in mineral transformer oil submitted to 50 Hz alternating overvoltages. Different shapes of streamer currents and electrical discharges have been recorded to develop a discrimination technique based on an Artificial Neural Network (ANN) and Wavelet analysis of these currents. This enables us to address a complementary diagnosis tool that can serve as an online transformer monitoring and protection.
© EDP Sciences, 2012