Short-Term Prediction of Offshore Wind Cluster Power Taking into Account Sensitive Meteorological Factors

To address strong fluctuations and uncertainties in offshore wind farm power output due to complex marine meteorology, this paper proposes a short-term forecasting method considering sensitive meteorological factors. It constructs spatiotemporal coupled correlation indices, combines DWI to quantify extreme weather effects, and uses a two-stage screening framework for sensitive factors. With IGRU as the base network and embedded physical constraints, it designs a composite optimization objective. Validation shows it identifies key factors, reduces false screening rate by 37%, and improves prediction accuracy by 23.5% (MAE 0.86, RMSE 1.15).