The output power of wind farms has significant randomness and variability, which results in adverse impacts on power system frequency stability. This paper extracts wind power fluctuation feature with the HHT (Hilbert-Huang Transform) method. Firstly, the original wind power data was decomposed into several IMFs (Intrinsic Mode Functions) and a tendency component by using the EMD (Empirical Mode Decomposition) method. Secondly, the instantaneous frequency of each IMF was calculated. On this basis, taking a WSCC 9-bus power system as benchmark, the impact on power system frequency caused by wind power fluctuation was simulated in a real-time simulation platform, and the key component which results in the frequency deviation was found. The simulation results validate the wind power fluctuation impacts on frequency deviation, underlying the following study on power system frequency stability under the situation of large-scale intermittent generation access into the grid.