Partial Discharge Detection By Classification of Tesla-Coil Music

Partial Discharge (PD) is an electrical leakage that crosses an insulation barrier under high voltage stress. The insulation (either solid or fluid) breaks due to other types of stress such as thermal or mechanical as well. Undetected PD leads to larger faults, arc flash and substantial damages to power systems assets. Over the past decades, different techniques have been developed for PD detection based on the nature, intensity, and location of such a fault. In general, external PD falls into two major categories: a) surface discharge, and b) corona. PD induces electromagnetic and audio responses that can be used for detection. This project explores spectrum analysis of audio signals that can be used to detect and classify audible PD (surface and corona). This exploration is performed on several Tesla-coil music as well as its normal instrumentation. Tesla-coil music may have features helpful in detecting PD especially when many real-world cases of PD are not readily available for analysis. The result of this work will be used in further algorithm development to classify and detect PD events in recorded audio signals.