Wind power is fluctuating and intermittent. Large scale wind power integration has an adverse impact on the power grid operation. This paper analyzes the characteristics of wind power fluctuations. It takes the wind power data as digital signal to process, based on digital signal processing. By means of analyzing autocorrelation coefficient, we find that fluctuation has a certain periodicity. Day periodicity component is extracted though finite Fourier decomposition. From the energy point of view, day periodicity component distribution is analyzed in different time scales and different installed capacity. The proportion of the periodic component changs with time and space of wind power. The longer the time period, the periodic component proportion is smaller. Along with the increase of the number of wind farms, periodic component proportion will increase signigicantly.