Pattern Recognition of Partial Discharges on Power Cable Systems

Abstract

Partial discharge measurements are considered to be an important diagnostic/condition monitoring tests on Power Cable System. Various off line and online measurement systems have come up with inductive and capacitive sensors. The high frequency partial discharge signals occur in power cable system due to various defects such as voids/cavity in the power cable insulation, defective termination, defective stress control materials and defective joints. However, the partial discharge pattern differs for each type of defects. Hence to identify the type of defect, proper analysis of pattern is required. The statistical parameters such as mean, skewness, Kurtosis etc. with respect to the phase angle, highest discharge magnitude etc. helps in extracting the feature information of each pattern. PD-fingerprints such as Skewness (Sk) and kurtosis which measures the degree of asymmetry of a distribution & sharpness of a distribution, along with the average value of each half cycle (Mean) are estimated using MATLAB programming for various partial discharge signals of laboratory failed Power Cables and accessories. In this work, an attempt is made to develop some finger prints for various defects on power cable systems using the statistical parameters and PD pattern.