Fault Detection Techniques for Power Transformers

Abstract

This paper presents the methodologies for incipient fault detection in Power transformers for off-line and on-line. An artificial neural network is used to detect off-line faults and whereas wavelet transforms are being used for on-line fault detection. The Dissolved Gas Analysis to detect incipient faults has been improved using artificial neural networks and is compared with Rogers ratio method with available samples of field information. The Wavelet transform techniques have been developed with different mother wavelets and their performances are compared. These have been used to detect incipient faults and also to distinguish between incipient fault and short circuit fault.