The invention belongs to the field of output forecast of renewable energy sources of a power system and particularly relates to an output forecast calculating method for the renewable energy sources by series compensation and parallel coupling of multiple forecasting models. When the renewable energy sources are forecasted, forecast precision of an existing forecast technology cannot meet actual demands of a project, and further an unscented Kalman filter (UKF) correction forecast method based on empirical mode decomposition (EMD) is provided. The method comprises the following steps: carryingout EMD on forecasting results of the multiple forecasting models to obtain multiple groups of stable time sequences; then updating and correcting by adopting a UKF method; finally, carrying out parallel coupling on the correction results by a least square method. The advantages of a certain model are amplified by weighing all the models, the influence of the defects of the model on the forecastprecision is reduced, and then the advantages of the multiple models are combined to obtain a high-precision wind power forecast result.