Abstract: In order to solve the problem of poor noise adaptability and mode aliasing in the extraction process of coupled sub/sup-synchronous oscillation parameters by common identification methods, an adaptive variational mode decomposition (VMD) is proposed in this paper. The total residual loss entropy, Chebyshev distance between components and edge entropy are defined to determine the number and bandwidth of decomposition mode, and the parameters of the decomposed oscillating components are identified by using the least squares and rotation invariant technique (TLS-ESPRIT), without additional noise reduction algorithms. The effectiveness of the proposed method is verified by composite signal tests and PSCAD/EMTDC simulation. Finally, the proposed method is compared with the improved Prony algorithm and MCEEMD under different noise levels and frequencies of sub/sup-synchronous oscillation. The results show that the proposed method can be adopted to effectively suppress the noise interference of the original signal, and is more accurate in the decomposition of the coupled sub/sup-synchronous oscillation signal, and the parameter identification results are more reliable. Hence, the conclusion has a certain reference significance for oscillation tracing and improvement system damping of a wind power system.