Identification of Partial Discharges in Gas-insulated Switchgear by Ultra-high-frequency Technique and Classification by Adopting Multi-class Support Vector Machines

Abstract

Ultra-high-frequency signals are generated due to particle movement, floating conductors, corona, and surface discharges in gas-insulated switchgear. The ultra-high-frequency signal generated due to particle movement is independent of operating pressure and applied voltage. The bandwidth of the ultra-high-frequency signal formed due to corona and surface discharges vary with applied voltage. In a ternary plot, each type of discharge has a unique location. If a variety of discharge occurs simultaneously, it is difficult to classify the type of discharges through the ternary plot. It is demonstrated that the type of discharge can be classified efficiently by adopting a multi-class support vector machine.