Wavelet and Mathematical Morphology as the de-noising methods for PD analysis of high voltage transformer windings

Abstract

Partial discharge (PD) analysis is one of the most important techniques to evaluate the condition of the insulation systems within high voltage (HV) transformers. However, in typical field environments, measurements of PD signals can be distorted by noise sources. This greatly reduces the ability to identify PD sources in HV transformer windings. Therefore, de-noising methods in PD analysis are very important. In recent years, several noise reduction techniques have been proposed for application in PD analysis. The common types of discharge events that may occur within high voltage transformer windings namely void, surface, corona and floating discharge have been experimentally generated. Each type of discharge was injected into different locations along a HV transformer winding and then measured using two wideband radio frequency current transformers (RFCTs) positioned at each end of the winding. Then, either the Discrete Wavelet Transform (DWT) and Stationary Wavelet Transform (SWT) or Mathematical Morphology (MM) were applied to reduce the noise in the raw captured PD signals. This paper presents the comparison of performance of the techniques in terms of noise reduction for this type of application.