Transformer differential protection using wavelet transform

Abstract

This paper will propose a cascade of minimum description length criterion with entropy approach along with artificial neural network (ANN) as an optimal feature extraction and selection tool for a wavelet packet transform based transformer differential protection. The proposed protection method provides a reliable and computationally efficient tool for distinguishing between internal faults and inrush currents. The role of minimum description length criterion with entropy approach has been found to improve the efficiency of ANN with the dimensionality reduction of the feature vector. This reduction plays a major role in preventing the redundancy effect that can occur when using several features in an intelligent based monitoring system.