Skip to Main Content
Almost all cases of insulation degradation in high voltage cables are due to partial discharge (PD) activity. To date, wavelet based analysis has been widely used to extract PD pulses from noisy environments. This paper explores the use of time domain features, namely short-time energy and short-time zero-crossing counts, to detect the presence of partial discharge signals prior to de-noising the signal for further investigation. In order to demonstrate the effectiveness of short-time energy and zero-crossing counts to identify PD signals embedded in noise, these features are tested with laboratory data. To further verify these results, real data was collected from a substation and the overall results demonstrate that these two time domain features are very effective in identifying PD pulses and are computationally efficient such that they can be considered for use in online PD monitoring.