In order to improve the accuracy of parameters identification and noise adaptability of sub-synchronous oscillation(SSO)for the doubly-fed induction generator(DFIG)

as well as to eliminate mode aliasing in identification

an SSO parameters identification method for DFIG based on the improved variational mode decomposition(VMD)and the total least squares-numerical subspace state space system identification(TLS-N4SID)was proposed. The grid-connected current of the DFIG was decomposed using VMD. Bayesian optimization(BO)was employed to refine the VMD procedure by determining the optimal number of intrinsic mode functions(IMFs)

K

and the approp

riate penalty parameter

α

thereby eliminating mode aliasing and improving noise adaptability. The IMFs produced by the decomposition were evaluated through mutual information(MI)analysis with the original grid-connected current signal

allowing the dominant IMFs to be identified. The dominant IMFs were resampled and subjected to parameters identification using TLS-N4SID. During this process

the N4SID was improved by a non-dominated sorting genetic algorithm II(NSGA-II)to determine the optimal signal subspace order

b

thereby enhancing identification accuracy and noise adaptability. The TLS method was employed to identify the characteristic parameters of the DFIG SSO signal. The proposed identification method was validated through tests on synthetic composite signal

time-domain simulations of a four-machine two-area system incorporating a DFIG-based wind farm

as well as real SSO records from the Guyuan wind farm in Hebei Province

all of which confirmed its effectiveness.