A Power Waveform Classification Method for Adaptive Synchrophasor Estimation

This paper designed a novel multiresolution analysis method and input waveform classification scheme. The proposed techniques serve as preprocessing procedures of input waveforms in order to facilitate synchrophasor estimation. Simple, scalable and time-shifted “pseudowavelets” (PWs) are employed to repeatedly calculate the correlation coefficients between the proposed PWs and input power waveforms. By scrutinizing the correlation factors with respect to frequency and time, the proposed method is capable of revealing the temporal trajectories of frequency and amplitude features of input waveforms. Such features are then leveraged to classify input waveform types. The result of such waveform classification can be applied to perform accurate synchrophasor estimation. Contrary to traditional approaches, where a single algorithm is designed to compute synchrophasor accurately for all input signal types, in this paper, a framework is proposed which enables adaptive switching of synchrophasor algorithms, so that the most suitable algorithm can be used for the identified waveform. The efficacy and efficiency of proposed methods are validated with standardized phasor measurement unit testing waveforms and simulated power system waveforms. The results prove that the PW-based waveform classification method is capable of distinguishing between power system dynamic waveforms.