For the subsynchronous oscillation phenomenon of the power system, this article proposes a modal parameter identification method with strong anti-noise and high recognition accuracy. The method uses the Fast independent component analysis algorithm to improve the autoregressive moving average (ARMA) algorithm. The FastICA algorithm can effectively separate the original signal from the noise. The FastICA algorithm has better anti-noise ability. First, the original signal is analyzed and preprocessed by the FastICA algorithm, and then the modal parameters of the original signal are identified according to the established ARMA model. The research results show that compared with the traditional ARMA algorithm identification results, this method can effectively improve the signal-to-noise ratio and provide more accurate identification results.