Improved non-linear least squares method for estimating the damping levels of electromechanical oscillations

Abstract

In this study, a new estimation algorithm is proposed for determining the damping of real-life electromechanical oscillations in power systems. The main contribution of this study is to enhance the inherent resilience of the methodology in estimating the damping of electromechanical modes, even in the case of potential data packet dropouts in time-synchronised acquired signals. In this regard, the basic idea was to exploit the inherent regularisation action of the Hilbert transform on the data packet dropouts. The secondary contribution relies on tailoring a rigorous method for estimating the damping levels, thus improving a recent procedure, proposed by the same authors, which was based on linear regression over a proper calculation interval. An improved least squares algorithm is hence tailored. It employs a specific objective function incorporating information regarding the finite Hilbert transform of the power signal components. The higher performances of the latter with respect to the others investigated are because of the regularisation action of the Hilbert transform, suitably integrated, in semianalytical manner, in an optimisation problem. The compact formulation of the estimation problem, characterised by a drastic reduction of the computational burden, allows a feasible implementation for real-time assessment in WAMS environments.