Abstract—This paper presents the use of Artificial Neural Networks (ANN) as a pattern classifier for the combined differential protection of generator-transformer unit with an aim to build a backup protection system to improve the overall reliability of the system. The proposed neural network model is trained and tested with an efficient Resilient Back propagation (RPROP) algorithm and Genetic Algorithm. The results are then compared. The neural network model makes the discrimination between operating conditions (like normal, magnetizing inrush, over-excitation conditions in transformer) and internal faults in transformer and generator based on the differential current waveform patterns. The proposed method is independent of amplitudes of the waveforms. Various normal and internal fault conditions of the transformer and generator are simulated using toolboxes in MATLAB/SIMULINK in order to obtain the differential current data used for the training and testing of the ANN.
 

Index Terms—artificial neural networks, differential protection, genetic algorithm, pattern recognition, resilient back propagation, unit protection

Cite: H. Balaga, D. N. Vishwakarma, and H. Nath, "Artificial Neural Network Based Backup Differential Protection of Generator-Transformer Unit," International Journal of Electronics and Electrical Engineering, Vol. 3, No. 6, pp. 482-487, December 2015. doi: 10.12720/ijeee.3.6.482-487