An Algorithm for Indentifying the Arrival Time of PD Pulses for PD Source Location

Abstract

In terms of locating the source of partial discharge, transient earth voltage (TEV) detection is well recognised to have advantages over electrical methods. A variety of sensors are suitable for on-line PD testing of outdoor equipment, including ultra-high frequency (UHF) and radio frequency (RF) antenna, transient earth voltage (TEV) sensors, acoustic emission (AE) sensors and high frequency current transformers (HFCT). Among these, UHF, TEV and AE are usually adopted to detect PD activity and to determine the partial discharge (PD) location. These three detection methods display similar response to PD activity. The main characteristic shown is multi-oscillation and gradual attenuation. To locate the source of a PD requires correlation of the PD pulse arrival times, accuracy of the location is determined by the ability to discriminate the difference in arrival time across the sensors. Unfortunately, since the UHF and TEV detection systems have higher sensitivity than others, the major problem faced in determining the arrival time of each pulse is the effect of interference and noise. The present paper proposes the use of discrete wavelet transform (DWT) to overcome the problems associated with noise in TEV measurement of PD in power plant. The method, which applies an effective denoising algorithm developed by the present authors to extract TEV signal from noise and identify the arrival time, is proved to be effective in on-site tests.