To address the issue of degradation in modal parameter identification accuracy caused by mode aliasing and noise contamination in power system disturbance recordings

a power oscillation modal parameter identification method based on enhanced variational mode decomposition (VMD) and Bayesian inference is proposed. First

a sparsity constraint is introduced into VMD

and a genetic algorithm is employed to optimize the number of modes and the penalty parameter

thereby improving the separation performance of oscillatory components. Then

a Bayesian inference model based on Markov chain Monte Carlo sampling is established. Moreover

the time-domain oscillation model is taken as the likelihood function

the posterior distributions of amplitude

frequency

damping factor

and phase are estimated

and the corresponding confidence intervals are obtained. Simulation cases and measured disturbance recordings from the Hunan power grid show that the proposed method can still be adopted to identify modal parameters accurately when the signal-to-noise ratio is no lower than 10 dB

while exhibiting good noise robustness and uncertainty quantification capability. The proposed method can provide supports for oscillation mechanism analysis and dynamic stability assessment in power systems.