Using a neural network for transformer protection

Abstract

A new method of using artificial neural networks (ANN) to identify the magnetizing inrush currents that may occur in transformers during start-up is developed in this paper. The method is based on the fact that magnetizing inrush current has large harmonic components. Using the backpropagation algorithm, a feedforward neural network (FFNN) has been trained to discriminate between transformer magnetizing inrush and no-inrush currents. The trained network was verified using test data from a laboratory transformer. Results presented in this paper indicate that the ANN based inrush detector is efficient with good performance and reliability.