Abstract:With the continuous expansion of new energy penetration and grid interconnection scale, low-frequency and ultra-low frequency oscillations appear in power grids with a high proportion of hydropower, showing strong nonlinear and non-stationary characteristics, which increases the difficulty of oscillation parameter identification. Aiming at the problem that the modal decomposition number K in Variational Mode Decomposition (VMD) cannot be determined adaptively and the problem that time-varying oscillations are prone to modal aliasing, a generalized variational mode decomposition (GVMD), combined with Teager-Kaiser Energy Operator (TKEO) for fast modal parameter identification of low-frequency and ultra-low-frequency oscillations. First, GVMD obtains the modal decomposition number K based on wavelet time-frequency analysis, and combines with generalized Fourier transform to make each mode clearly defined in the time-frequency plane. Then, after the signal is decomposed by VMD, the TKEO method is used to quickly complete the identification of oscillation parameters. Finally, the effectiveness and feasibility of the proposed method are verified by self-synthesized signal, EPRI-36 system simulation and measured power grid signal simulation.