Power spectral density estimation of randomly sampled partial discharge signals

Results of investigations performed in order to verify whether the Power Spectral Density Function (PSDF), estimated from a sample of Partial Discharge (PD) pulse signals, can summarize the spectral characteristics of the whole population of discharge pulses generated by a PD phenomenon, are reported. The PSDF of sequences of PD pulses produced by either corona or surface discharges, taken as a reference, are compared with those obtained from samples of each population, selected using the Poisson technique. It is shown that an accurate PSDF is obtained already with a relatively-small amount of pulses, particularly for corona discharges, and that the PSDF of the two PD phenomena differ at high frequencies.