Ultra-short-term wind power ramp events have been increasingly influencing the wind machine operation in power systems. The domestic definition of ramp events is not clear and the corresponding forecasting methods are absent at present. The physical meaning of the wind power ramp events is elaborated and a combination forecast method based on the atomic sparse decomposition (ASD) and back propagation (BP) neural networks is proposed. Atomic components self-prediction model

error component prediction model and the combination prediction model are established separately. The different prediction methods and different time space measured data are analyzed comprehensively through applications with real wind farm data. The results show that the proposed method can improve the prediction accuracy

and the statistical intervals of absolute mean error and root mean square error are also remarkably reduced.