Differential Protection of Indirect Symmetrical Phase Shift Transformer and Internal Faults Classification using Wavelet and ANN

This paper illustrates a differential protection algorithm for indirect symmetrical phase shifting transformer (ISPST) using wavelet transform (WT). Further, a Multi-Layer Feed Forward Neural Network (MLFFNN) based algorithm has been developed for classification of internal fault in ISPST. Detailed coefficient at level four (D4) of phase current is used as input vector for MLFFN network. Principle component analysis (PCA) at input reduces the burden and makes the detection and classification algorithm fast. Genetic Algorithm (GA) is used to obtain the optimal structure of MLFFNN. The discrimination between internal fault and magnetizing inrush is developed based on the time elapsed between the instant of inception of disturbance and the instant of the maximum peak in frequency component D4 of WT. It distinguishes magnetizing inrush and internal fault within quarter cycle after disturbance. An ISPST is simulated using PSCAD/EMTDC and RSCAD/RTDS platform to obtain the differential current signal.