Diagnostic evaluations of electrical apparatus by a novel partial discharge measurement system

The application of a novel partial discharge measurement system to the identification of defects in insulation systems of electrical apparatus is presented in this paper. It records the whole waveshape of a large number of partial discharge pulses and, then, splits the measured dataset into subgroups on the basis of the pulse waveshape. When different waveshapes can be attributed to different partial discharge sources or, possibly, noise, artificial intelligence techniques enable the system to reject noise and apply identification/risk-assessment routines to homogeneous patterns, that is, to phase-resolved partial discharge patterns relevant to a single partial discharge source.