Application of electromagnetic vector sensor for partial discharge on-line monitoring of transformer

Partial discharge (PD) sources identification and orientation is one of the key problems for PD online monitoring of transformers. Electromagnetic vector sensor (EMVS) has been successfully applied to estimate the electromagnetic sources spatial parameters and polarizations. In this paper, a single EMVS is proposed to measure the ultra-high frequency (UHF) signal for PD detection and orientation. With its output data processed by a narrow-band filter, the number of PD sources can be estimated exactly by using minimal describing length (MDL) rule or smoothed rank sequence (SRS) method, based on which, the spatial parameters and polarizations of PD sources can be obtained by adopting MUSIC algorithm to estimate the spatial spectrum of the data's covariance matrix. The results of simulate experiment show that this method is capable of identifying three PD sources and estimating their azimuth, elevation and polarization. It is practical and hopeful to realize the PD orientation in transformer online PD monitoring.