Off-line testing of a wavelet packet-based algorithm for discriminating inrush current in three-phase power transformers

This paper introduces a novel current diagnosis and protection technique, which is based on wavelet packet transform (WPT). The WPT algorithm is implemented and tested, where the output of the second level details is able to provide the needed signature to diagnose the type of the current flowing through the transformer. The proposed technique is implemented using Daubechies Mother wavelet, and simulated using MATLAB on differential currents data collected form a prototype three-phase power transformer. The WPT technique performed successfully by identifying different currents data including magnetizing inrush, normal current and different fault currents for different types of loading and energizing conditions.