Defect classification based on Weibull statistic of partial discharge height distribution with wavelet preprocessing

The paper describes a stochastic analysis performed using two-parameter Weibull statistics involving a scheme whereby the PD signals are decomposed using discrete wavelet transform. In a high voltage experimental setup, artificial cylindrical voids were made in a Perspex column of 40 mm in length. The signals obtained from a corona detector were collected via an 8 bit oscilloscope with data storage capabilities. The signals were analyzed using PD height distribution (PDHD). The scheme presents itself as a multi-level Weibull analysis to identify and quantify voids in solid dielectric insulations based on nonultra wide band detection and wavelet aided signal processing.