Risk functions for partial discharge inference in insulation systems

Partial discharges taking place in insulation systems may have different natures and locations. The identification of features which can be correlated to the partial discharge (PD) phenomena occurring in the system under test is a fundamental requirement for insulation diagnosis. The identification problem may be of difficult solution when more than one phenomenon are simultaneously occurring. In this paper, two-parameter Weibull distributions and multiplicatively-mixed Weibull distributions are used to infer data from PD measurements performed on specimens of stator bars laboratory-aged under accelerated procedures