Robust ensemble data analytics for incomplete PMU measurements-based power system stability assessment

Abstract

This letter proposes a new ensemble data-analytics model for PMU-based pre-contingency stability assessment (SA) considering incomplete data measurements. The model consists of a minimum number of single classifiers which are, respectively, trained by a strategically selected cluster of PMU measurements. Under any PMU missing scenario, the power grid observability from available PMUs can still be ensured to the maximum extent to maintain the SA accuracy. The proposed method is verified through both theoretical proof and numerical simulations.