Unbiased forecasting of seasonal wind power generation based on a novel seasonal multivariable grey model

Accurate short-term wind power forecasting is crucial for enhancing the operational efficiency and grid integration of wind farms, contributing significantly to pollution reduction and energy conservation. To address the seasonal variations and multi-factor dependencies inherent in wind power generation, this study develops a novel seasonal multivariable grey model (SMGM(1,N)) incorporating numerical weather prediction data. The proposed SMGM(1,N) is proven to provide unbiased predictions for short-term wind power generation. Based on this model, the interval prediction is designed using an intelligent optimization algorithm and the Bootstrap method. For illustration and verification purposes, Belgian onshore and offshore wind farm generation sets are studied. Results demonstrate that SMGM(1,N) achieves superior forecasting accuracy, yielding the lowest MAPE values of 1.74 % (onshore) and 1.76 % (offshore) in point prediction, alongside optimal performance in interval prediction metrics (coverage width-based criterion and average interval score) compared to six benchmark models. This advancement provides reliable generation uncertainty quantification for improved wind farm management.