A novel two-step framework for day-ahead wind power probabilistic forecasting considering uncertainties and ramp events

作者：Yang Cui, Cheng Chi, Yan Zhu, Dan Meng, Zhenghong Chen · 发表于：International Journal of Electrical Power & Energy Systems · 年份：2025 · DOI：10.1016/j.ijepes.2025.110780 · 被引用次数：9 · 研究领域：Energy Load and Power Forecasting、Electric Power System Optimization、Integrated Energy Systems Optimization

The large-scale integration of wind power into power grids can lead to substantial challenges to its safety and stability, which makes wind power probabilistic forecasting (WPPF) increasingly critical for power system operations due to its ability to capture the uncertainties associated with prediction errors. However, wind power ramp events (WPREs) can significantly reduce WPPF accuracy and potentially cause severe grid disturbances. To address this issue, this study proposes a novel LSTM-WPRE-PF model for day-ahead probabilistic forecasting by incorporating deterministic forecast with WPREs. First, the uncertainties in wind power generation were analyzed, and the correlations between key meteorological factors and WPREs were explored. Second, we quantified the uncertainties of wind power fluctuations using Wavelet Packet Variance Entropy (WPVE) and examined the impact of WPREs. Third, we compared four deterministic wind power forecasting models and selected the optimal one, LSTM-WPRE, as the foundation of the proposed two-step model. Finally, the LSTM-WPRE-PF model was evaluated against four models using real data from two mountainous wind farms in Hubei, China. Results show that LSTM-WPRE-PF outperformed the other models in terms of reliability, sharpness, and composite evaluation metrics. These findings highlight the model’s potential to enhance the safety and economic efficiency of power system operations.