Using Fuzzy ARTmap neural network for determination of partial discharge location in power transformers

Abstract

Techniques for locating a partial discharge source are of major importance in both the maintenance and repair of a transformer. This paper presents a novel approach to identify partial discharge locations in transformer winding using neural network. In this paper for simulation and detection of partial discharge, detail model of transformer is used. With modeling of partial discharge impulse source in EMTP software, this phenomenon is implemented in different points of transformer winding. Then produced current in both ends of winding is measured and use for training and test of neural network. In actual, obtained current signals is with noise. Thus in this paper the performance of the Fuzzy ARTmap neural network for correct determination of partial discharge location in power transformer with considering different noises on simulated current signals for simulation of actual conditions is surveyed. The most important characteristics of neural networks are capabilities to learning and predict the various patterns and other is capability to provide a fast responsible for input patterns. The neural network used here for simulation patterns trainings and testing of the partial discharge in power transformer winding is Fuzzy ARTmap.