Abstract: Based on the suppression of grid-connected sub-synchronous oscillation of doubly fed induction generator. In order to reduce the errors in the modal identification of sub-synchronous oscillation and improve the anti-noise performance of Prony algorithm,this paper proposes a modal identification method of sub-synchronous oscillation based on empirical mode decomposition（EMD）and TLSProny algorithm and a voltage feedback suppression strategy. Firstly,EMD is used to preprocess the oscillation signal to reduce the impact of noise on the identification of oscillation modes by Prony algorithm. Then,the TLS-Prony algorithm was used to further suppress the noise interference,extract the oscillation features,and identify the oscillation modes. Finally,the output voltage of the grid-side converter is modulated directly by adjusting the control parameters,and the equivalent positive resistance is injected into the system to improve the negative damping characteristics of the system. An additional damping controller is designed with the cascade capacitor voltage as the feedback input signal. The influence of additional damping on the stability of the system is given by modal analysis,and the effectiveness of the proposed damping suppression strategy is verified by Matlab/Simulink simulation platform.