The wind powerhas the qualifications of random and volatility concentration.This paper present a combined prediction based on ensemble empirical mode decomposition (EEMD) and autoregressive conditional heteroskedasticity model(ARCH).By means of the EEMD,the wind power sequence is decomposed into a series of stationary component.Then the components are reconstructed into fluctuant components,medium-term trend and long-term trend components for centralizing the characteristic information and reducing the difficulty of predicting.Finally,considering different types of components,the different ARCH model is built.Simulation results show that the combined prediction can offer more accurate forecasting result and reflect the fluctuation of wind power.