"A transformer is one of the essential parts of the power systemnetwork. Since the power transformer outage can cause great blackouts, its protection is one of the challenging subjects in the area of power system protection. The crucial objective of protective system network is to reduce the mal-operation and enhance the reliability of electric energy flow in power system network.The conventionalDifferential protection is used to protect transformer frominternal faults and restrains during inrush currents and external faults, with the help of restraining coil. The restraint signal reduces the sensitivity of the uses relay. Hence there is need to investigate in to new methods which coil restrain the relay from external fault and inrush current with advanced signal processing technique.The proposed research work aims at using Wavelet transform based multi-resolution analysis for transformer protection. For the analysis, three phase current signals are measured and sampled from the primary winding of the transformer. These signals are decomposed with db2 mother wavelet to obtain approximate coefficients using wavelet transform. A cumulative summation index is obtained by summation these approximate coefficients over a moving window of half cycle. This index is compared with the threshold for fault detection and classification. The proposed algorithm was found to be successful in the detecting and classifying the internal fault and discriminate the other transients associated with external fault and inrush current.
"