A Novel Digital Image Processing Technique to Detect Power Transformers Internal Faults using Frequency Response Analysis Polar Plot Signature

Abstract

This thesis introduces a new application, which aims to improve the current approach to interpretation of the Frequency Response Analysis (FRA) signature. In contrast to current FRA practice (conventional approach), the newly proposed application develops various Digital Image Processing techniques to overcome most of the drawbacks of the conventional FRA approach. By automating and standardising FRA interpretation, its detection accuracy is enhanced so that incipient mechanical and non-mechanical faults within power transformers can be detected.