GLRT-Based Mitigation of Partial Discharge Effect on Detection of Radial Deformation of Transformer HV Winding Using SAR Imaging Method

Abstract

Online and on time detection of defects leads to technical and economic merits for networks. Recently, an online method has been introduced for the radial deformation detection of transformer high-voltage winding using electromagnetic waves called an ultrawideband synthetic aperture radar (SAR) imaging method. The basis of this method is comparison of sound and deformed images of transformer windings to detect the radial deformation. Partial discharge (PD), as another defect in the transformers, emits ultrahigh-frequency (UHF) signals that contain 0.3 to 3 GHz frequencies. PD detection by using UHF antennas has been tested on transformers. If an SAR imaging method can be applied in UHF band, a set of antennas can be used for the detection of both defects and get more information about transformer condition. In this paper, a feasibility study into the application of UHF band in an SAR imaging method is carried out. It is shown that PD, during a UHF SAR imaging process, leads to image distortion and wrong decision about deformation condition. In order to achieve correct imaging, a UHF stepped-frequency imaging method is tested for the detection of radial deformation and generalized likelihood ratio test (GLRT) is adapted and applied to solve the problem of the PD effect on imaging process. In addition, the effects of GLRT parameter variations on the problem are investigated in this paper.