Localization of short circuit faults in transformers using harmonic analysis and pattern recognition

Abstract

The power transformers are among the most important and expensive equipments in the electrical power systems. So the transformer protection is an essential part of the system maintenance and protection. This paper presents a new method for locating transformer winding faults specially short circuit faults such as turn-to-turn, turn-to-core, turn-to-earth, turn-to-transformer body, and primary winding to secondary winding. In this study a dynamic model of transformer is simulated. The current and voltage signals of input and output terminals of the transformer are measured for different kinds of faults. The Fourier transform of all of the measured signals are calculated. Then the harmonics of all of the simulated faults are extracted and saved as special indexes in a database and the fault locating process is completed by the pattern recognition. And in the last step, the pattern recognition determines the fault's location.