Multiple Partial Discharge Signal Decomposition using Mathematical Morphology

Abstract

Partial discharge (PD) measurements are an important technique for assessing the condition of power equipment especially in high voltage (HV) transformers. Different PD sources may have different effects on the condition and performance of power equipment insulation. Therefore, identification of PD sources is a great interest for both system, utilities and equipment manufacturers. An experiment has been designed to access the methodologies for location of multiple PD sources within a high voltage transformer winding. It is assumed that, the response is attenuated and distorted by the propagation path taken and termination characteristics altering the output waveforms during the propagation of the signals along transformer windings. This produces changes in the energy characteristics of the signals when they reach both measurement sensors. In order to analyse the measured data and to produce energy vectors associated with the signals, signal decomposition techniques are required. The purpose of signal decomposition is to reveal intrinsic components which are representative of the measured PD pulses related to the corresponding PD sources in decomposed signals. Two decomposition techniques namely Wavelet analysis and Mathematical Morphology were used in order to decompose the measured PD signals from both measurement points.