Data-Driven Modal Features Extraction Through the Variational Mode Decomposition Method

This paper presents a low-frequency power system modal features extraction method for exploiting the Variational Mode Decomposition (VMD) method. The features extraction process is focused on a time-frequency analysis of nonlinear and non-stationary signals resulting from a large disturbance. The insight from using VMD is that the technique can accurately achieve signal decomposition in obtaining sub-band signals or modes from power oscillating signals. An optimal and recursive time-frequency formulation yields the modes and their instantaneous modal features such as amplitude, frequency, damping and energy, by applying Hilbert transform. The results demonstrate the applicability of the proposition in a large-scale power system.