Due to the long-term effects of human activities and climate change, the hydrological time series exhibit more complicated variation features such as multiple time scales,many ruense dynamic change and self memory function,which increase the uncertainty of hydrological forecast result.The pape established a new hydrologic forecast model based on the empirical mode decomposition model, kernel principal component analysis model and support vector machine model,and choose Nash efficiency, self correlation coefficient, relative error as the multi objective criteria of forecasting precision and parameter calibration. The model was applied to the long-term runoff series at Huayuankou hydrology station of the Yellow River.The results show that the forecast time of the model is long and has better accuracy and practical value. The model can provide a method for the prediction of complex hydrological time series of multiple factors effect.