Fault Classification of Dry Type Transformer Using Pattern Recognition Neural Network

Abstract

IoT (Internet of Things) has revolutionize the working of the machines in the modern industry. Different electrical power equipment’s are being monitored used using IoT. Dry type transformers (DTT) applications are limited therefore are not used widely in distribution due to its high manufacturing cost. Although they are fireproof, required convenient maintenance but fail due to windings insulation failure and overloading. Nowadays Artificial Intelligence (AI) systems with IoT are widely used in monitoring of transformers with great success. Artificial Neural Network (ANN) used for the identification and diagnosis of the fault in oil type transformers. In ANN systems, supervised learning is often used in which models are trained with datasets such a way that they are able to classify the presence or absence of faults in the transformer under testing. In this research paper, pattern recognition-based method for detecting various DTT failure states is presented. In this suggested technique the real time parameters of the DTT were collected, normalized and then used as an ANN inputs. Four different classes of DTT conditions: healthy, open circuit, short circuit and overload faults were taken into consideration. The outcomes shows that the suggested approach is extremely successful in diagnosing various DTT defects with a high percentage of success.