ANN based Technique for Discrimination between Magnetizing Inrush and Internal Fault Currents in Transformer

Mudita Mathur, Vinay Barhate

Abstract: This paper presents the application of artificial neural network based pattern recognition approach for differential protection of transformer. The power transformer protective relay should block the tripping during magnetizing inrush and rapidly initiate the tripping during short circuit. The ANN based algorithm is designed and trained with experimental data in laboratory using single phase Transformer, which can be extended to Three phase power Transformers, for sinusoidal wave which permits the relay operation only during short circuit conditions and not for inrush current. Solution to Backpropagation Algorithm is given using C programming language. Modeling of ANN using feed forward Backpropagation techniques is introduced. Voltage signals proportional to current is derived through CTs. Data Acquisition is done using ESA interface card at a particular sampling rate using PC.

Keywords: Artificial neural network, Pattern recognition, Transformer protection, Protective relay, Inrush current