The online identification of subsynchronous oscillation modal parameter became possible with a wide use of WAMS in power system. However, there is much power electronic equipment in the power system, which causes strong interference noises in WAMS sampled signal, and the veracity of vibration modal parameters is influenced. Because of the Fast Independent Component Analysis (FastICA) could separate noise signal from observed signal. This paper first presents to pretreat sampling signal through the FastICA. And then it identifies the denoised signals through Matrix Pencil (MP) to obtain the oscillation modal parameter, with which he recognition accuracy of MP could be enhanced based on FastICA-MP. This paper takes an ideal example and a UHVDC system in China as the simulation examples. The simulation results show that the FastICA can separate noise signal effectively and improve the MP identification accuracy, laying the foundations for the design of supplementary subsynchronous damping controller.