PD pulse distribution pattern analysis

Statistical data analysis has been requisite for partial discharge (PD) study. This has been enabled by the development of microprocessors or the downsizing of computers. It is important to pay much more attention to profiles of a packet of PD pulses on the voltage phase angle. Statistical parameters such as the skewness and the kurtosis have been intensively investigated for PD analysis. More recently, the neural network algorithm has been utilized especially to discriminate PD signals from noise signals for insulation diagnosis.<<ETX>>