Location of Disk Space Variations in Transformer Winding using Convolutional Neural Networks

Abstract

In this paper, disk space variations (DSV) as one of common transformer winding defects, has been practically applied to a transformer winding in some specific locations and with various extents. To locate DSV faults, Convolutional neural networks (CNN) has been applied to frequency response traces of the tested winding. It has been presented that the proposed method has accurate fault location capability. Convolutional Neural Networks was utilized to extract important features from frequency response traces to detect DSV location in transformer winding.