Time Series Classification for Locating Forced Oscillation Sources

This article presents a machine learning based time-series classification method for using synchrophasor measurements to locate the source of forced oscillation (FO) fast disturbance removal. First, multivariate time series (MTS) matrices are constructed by most informative selected sequential feature selection from each power plant. Then, Mahalanobis matrix is trained such that distance between MTSs same class (i.e., with FO location) minimized and different classes locations) maximized. allows be classified classifiers membership corresponding location source. To meet runtime requirements online matching, templates reduce data size improve matching efficiency. account uncertainty in identifying exact beginning an event, dynamic warping used align out-of-sync MTSs. IEEE 39bus WECC 179bus systems algorithm development validation. Simulation results demonstrate meets operation requirement high accuracy misaligned sets.