Signal Processing and Classification of Synchro-phasor data

The study in this paper presents post-processing of synchrophasor data to analyze the time-frequency characteristic variation of signals obtained at different buses in the power network. The time-frequency representation approach is applied to analyze the pulsations that contain rapid variations in amplitude or phase during the event occurrence against preevent/post-event conditions. The developed procedure gives reasonable results for highlighting the distinct representation of events for the non-stationary signals considered in study. This is based on reassigned smoothed pseudo-Wigner-Ville distribution (RSWVD). Further, classifier using support vector machines (SVMs) is applied to create discriminant patterns into predefined classes between event and post-event conditions. The performance analysis shows high classification accuracy.