Due to the prevalent characteristics of “small data,” “seasonality,” and “periodicity” in China's renewable energy power generation data, there are significant challenges in making long-term power generation forecasts. Consequently, this paper employs a data preprocessing method involving periodic aggregation to enhance the “quasi-exponential” characteristics of the original data, mitigate the effects of “seasonality” and “periodicity,” and subsequently apply the DGM (1,1) model for forecasting the aggregated data. The model method uses simple data preprocessing and traditional grey prediction model to achieve accurate prediction of nonlinear period data, and to a certain extent overcomes the disadvantages of complex model structure and difficult technical implementation compared with the traditional seasonal grey prediction model. To validate the accuracy and effectiveness of the model, a comprehensive comparison analysis is conducted using various prediction models with the China wind power generation dataset. The findings of the study demonstrate that the model outperforms other models in terms of predictive performance on both test and prediction sets.