Automatically Discerning Power System Dynamics in Synchrophasor Measurements Data Spectra

Techniques for time-frequency analysis can be used to identify, isolate, and track the emergence of numerous underlying power system dynamic processes just from measurements. These are especially useful for drawing meaningful conclusions from data obtained under ambient conditions, when power system phenomena can be less obvious in the time domain. However, because spectrograms have typically been evaluated by visual inspection, it is impossible to routinely perform assessment on large amounts of data. More crucially, relying just on visual inspection could result in missing crucial dynamics due to human error, e.g., setting incorrect color scaling. To address these issues, this work proposes an approach to automatically identify regions of interest on spectrograms over large data sets, which reflect the underlying power system’s dynamic behavior. Results for actual synchrophasor measurements from the Dominion Energy system are obtained to show the effectiveness of the proposed method.