Improved statistical up-scaling clustering wind power forecasting method based on spatial-temporal data enhancement

Uncertain local-global data mapping relationships in Wind Power Forecasting (WPF) models for wind farm clusters can be effectively addressed by combining Numerical Weather Prediction (NWP) with Wind Farm Actual Meteorological (WFAM). This approach improves prediction systems’ accuracy and stability. For wind farm clusters with better spatial-temporal mapping linkages simultaneously driven by NWP and WFAM, we present a unique improved statistical upscaling clustering (ISUC)-based WPF approach in this study. First, the NWP-sensitive meteorological parameters and wind power uncertainty are examined. Both NWP and WFAM data are used to construct a dynamic clustering technique. Typically, a number of nearby periods are used to dynamically choose wind farms. Second, to strengthen the mapping link between NWP and WFAM, a Spatial-Temporal Data Enhancement (STDE) model is created. STDE improves the forecast model’s capacity to adjust to local and worldwide weather variations. Lastly, a multi-objective optimization method is used to choose the prediction model parameter. We illustrate the accuracy and stability of the suggested approach using real-world operational data from wind farm clusters in North China. The findings demonstrate that this innovative strategy offers a new, high-precision technique for WPF in wind farm clusters.