Application of back-propagation neural network for transformer differential protection schemes part 2 identification the phase with fault appearance in power transformer

In this paper, a decision algorithm for identifying the phase with fault appearance of a two-winding three-phase transformer has been proposed. A decision algorithm based on a combination of Discrete Wavelet Transforms and back-propagation neural networks (BPNN) is developed. Daubechies4 (db4) is employed as mother wavelet in order to decompose high frequency components from fault signals. The maximum coefficients of DWT at ¼ cycle of phase A, B, C and zero sequence for post-fault differential current are used as input patterns for training process, and the results obtained from the decision algorithm are investigated. Various cases and fault types are studied to verify the validity of the algorithm. The result is found that the proposed decision algorithm can give more satisfactory results.