Abstract:To measure the aerodynamic noise of a 2.2 MW horizontal-axis wind turbine, six measurement points were set up in a large wind farm. The points were placed in the upwind, downwind, and at a 60° angle around the turbine. The trends of total sound pressure level (TSPL) of measurement points were obtained by 1/3 octave analysis under the wind speeds ranging from 6.55-15.00 m/s. On the basis, the empirical modal decomposition and Hilbert transform, combined with the marginal spectra, are applied to obtain the key time-frequency characteristics of wind turbine noise. The results show that the total sound pressure levels at the 60° measurement points are similar to those of the standard measurement points in the same direction. However, the noise level at measurement point #4, located upwind of the wind turbine, is relatively lower. The variance contribution ratio and correlation coefficient derived from the empirical modal decomposition reveal that the primary energy of the noise is concentrated in the frequency band of 770~1120 Hz. The principal frequency of the fitted marginal spectrum is in good agreement with the cubic relationship of the corresponding wind speed.