Power transformer protection using wavelet packet transform

Abstract

This paper presents a novel technique for power transformer protection. This technique is based on extracting current signals under different operating conditions such as during energization of transformer and different fault conditions. Fast wavelet transform (FWT) schemes may not be so effective for data that has mainly oscillating features. An effective solution to distinguish inrush current from internal fault cases involves examining the signals in both the time and frequency domains simultaneously. The wavelet packet transform is the extension of the FWT which allows fine characterization of signal content for both time and frequency together. In the proposed work A 138/13.8 kV transformer was simulated using Matlab simulink environment. The transient signals of differential currents are captured and analyzed using `db1' wavelet packet transform at level 2. The energy content of the second level approximation at different fault incidence angles is calculated. A threshold is set based upon number of simulation studies. By considering the threshold value as the reference, the magnetizing inrush current is distinguished from internal fault current. The results indicate that the proposed scheme of protection is stable, reliable, and fast during the discrimination between magnetizing inrush currents, and internal faults.