Transformer Differential Protection Based on Wavelet and Neural Network

This paper presents a novel power transformer differential protection scheme by using combined Wavelet Transform and Artificial Neural Network which provide the means to enhance the classical protection principles and facilitate faster, more secure and dependable differential protection for power transformer. Wavelet transform is used to extract the feature from transient signal and the neural network is trained by the extracted features of the transient signal to accurately discriminate between the internal fault and magnetizing inrush current. The wavelet transform is firstly applied to decompose the differential current of power transformer in to a series of detailed wavelet components and then the spectral energies of the detailed wavelet components are calculated. The obtained spectral energies are employed to train the Optimal Feed Forward Back propagation Neural Network (OFFBNN). A three phase 315 MVA, 220/400 kV, 50Hz, power transformer is modelled in PSCAD/EMTDC software and the algorithm is evaluated in MATLAB. The results clearly shows that the proposed scheme is reliable, accurate and fast than the conventional differential protection scheme.