Utilization of Resilient Back Propagation Algorithm and Discrete Wavelet Transform for the Differential Protection of Three Phase Power Transformer

Transformer is the most essential and costly equipment of the power system. Its protection against internal fault is achieved by implementing differential protection scheme. In the proposed work, Artificial Neural Network (ANN) and Discrete Wavelet Transform (DWT) are utilised for discriminating internal fault current from that of the inrush current. DWT is applied to extract the features by decomposing the current signal into series of frequency bands. Further, the extracted features of DWT are supplied to the Multi-layer Feed Forward Neural Network (MLFFNN) for classifying inrush and internal fault current. Initially, MLFFNN is trained by Resilient Back Propagation Algorithm (RBPA). Later, the same is trained with the help of most widely used Back Propagation Algorithm (BPA). Then, corresponding results are compared to examine the capabilities of RBPA. From the results, it is realized that RBPA is accurate and faster as compared to the widely accepted BPA.