Publication: Arcing Faults Detection in Switchgear with Extreme Learning Machine
Date
2022
Authors
Ishak S.
Koh S.P.
Tan J.D.
Tiong S.K.
Chen C.P.
Yaw C.T.
Journal Title
Journal ISSN
Volume Title
Publisher
Institute of Physics
Abstract
The robustness of switchgears has critical impacts on the general efficiency of power distribution systems. Faulty switchgears lead to many unwanted complications for utility bodies, which in turn lead to even bigger issues. In this paper, a remote arcing fault sensing technique is proposed using ELM. By analysing the sonic waves emitted, the proposed method is capable to detect possible arcing faults in switchgears. Tests and experiments have been conducted to investigate the performance of the proposed algorithm in detecting these arcing faults. The obtained results are analysed in time and frequency domains. In the time domain analysis, the results show 93.75% success rate in training stage, 95.83% in validation stage, and 87.5% in testing stage. In the frequency domain analysis, the results show 93.75% success rate in training stage, 91.67% in validation stage, and 100% success rate in testing stage. It is thus concluded that the proposed algorithm is capable to identify arcing faults in switchgears. � Published under licence by IOP Publishing Ltd.
Description
Frequency domain analysis; Knowledge acquisition; Time domain analysis; Arcing; Arcing faults; Extreme learning machine; Fault sensing; Faults detection; General efficiencies; Learning machines; Power-distribution system; Switchgear fault; Validation stages; Fault detection