Partial Discharge Detection and Diagnostics Using Double-Sided Measurement

Abstract

Partial discharge detection and diagnostics have always been the major concerns of power cable quality evaluation. Based on the review of published literatures and the systems currently available on market, this ME thesis has proposed a new mathematical algorithm for double-sided PD measurement, which removes earth noise from the collected PD signals and localises the PD sources. This developed algorithm generally is comprised of wavelet de-noising, time domain reflectometry, time-of-arrival estimation and very low frequency cable testing criteria. It has been implemented on LabVIEW platform. PD signals are acquired by USB-5133 and then process through the algorithm. This system has a graphical user interface which is able to give users results of the de-noised signals from two cable ends and readings about the PD source location. This system has been tested at both signal and medium voltage levels. Signal level tests give results as expected that it has the minimum errors when the PDs occur near the centre of the cable and vice versa. It performs well for the PD signals with signal-to-noise ratio no less than 9.70 dB. Medium voltage level tests do not give sufficient results due to radiation phenomenon between two cable ends. However, by analysing the waveforms, it partially verifies the integrity of the developed system.