Accurate and comprehensive understanding of the wind power output characteristics is the precondition of efficient utilization of wind energy resources, wind power output, however, due to the affected by the natural wind to exist severe volatility and randomness. For this, the wind power output characteristic is described first in this article, and then an adaptive data decomposition method is introduced, that is Ensemble Empirical Mode Decomposition (EEMD), the nonlinear and unstable active output of time series data is decomposed into several intrinsic mode functions (IMF) by EEMD, by observing the decomposition of the IMF and the volatility of each component to deeply understand the wind power output characteristics of sampling area, hoping it will be a new thought that better and more efficient use of large scale wind power in the future.