Wind power forecasting becomes a research hotspot gradually. To improve accuracy of forecast, the model becomes increasingly complex. This paper present a simple model for ultra -short -term wind forecasting by researching the wind power characteristics .Firstly, The time series of wind power is decomposed into trend component and fluctuant components. Considering different characteristics of each component, ARMA and ELM model is built individually. Simulation results show that this model is preferably fitting for the time series of wind power and has the good application prospect.