To realize the fast and accurate identification of sub/supersynchronous oscillations in power systems, this paper proposes a method based on the combination of multivariate synchrosqueezing wavelet transform(MSST) and sparse time-domain method (STD) to identify sub/supersynchronous oscillations.. The method firstly constructs the multi-channel wide-area measurement information matrix of the power system with nodes as units; then MSST is used to perform time-frequency transformation on the measurement information matrix to obtain the MSST coefficient matrix of each node; furthermore, the MSST coefficient matrix characterizing the sub/supersynchronous oscillation modes of the system is screened by using the improved ridge extraction method. Based on this, the time-domain reconstruction of the oscillation components is carried out, and then the STD time-domain modal identification method is utilized to identify the parameters of each mode, such as the frequency and damping ratio; finally, the accuracy and effectiveness of the proposed method are verified by the self-synthesized signals, the simulation signals of the grid-connected system of doubly-fed wind farms via string-complementary and the measured data of an actual wind farm.