Compressed-DMD for Electromechanical Mode Estimation from High-Dimensional Ambient Data

Abstract

Accurate and rapid estimation of electromechanical mode plays an important role in sensing the security situation of power systems. In this paper, the Compressed Dynamic Mode Decompensation (Compressed-DMD) based estimation approach was proposed to extract the electromechanical mode from high-dimensional ambient data measured by the synchrophasor measurement unit. To improve the efficiency of DMD in processing high-dimensional ambient data under the premise of ensuring calculation accuracy, the Compressed-DMD was introduced to generate the approximation of the high-dimensional left and right singular vectors by employing the aggressive random test matrices and truncated eigendecomposition. Simulation examples of IEEE 16-generator 5-area system and real measurements verify the feasibility and effectiveness of the proposed method.