Comparative Study on Different Wind Power Prediction Algorithms Based on Comprehensive Performance Evaluation

Considering that existing studies on wind power prediction performance evaluation struggle to fully capture the prediction performance, this paper constructs a comprehensive evaluation scheme that incorporates fundamental indices, long time-scale indices, and key time-period indices. Subsequently, weights are automatically assigned to these indices through the entropy weighting method, and the weighted sum is calculated to derive the final comprehensive score for each prediction result, enabling a scientific and comprehensive evaluation of wind power prediction performance across various time scales. Comparative studies are conducted using measured data from a wind farm in China and predicted power data from various vendors. The results demonstrate that the comprehensive evaluation scheme enables a scientific and holistic assessment of wind farm power prediction results. Furthermore, an analysis of the error characteristics is conducted, revealing potential limitations of the wind power prediction algorithms deployed by actual vendors.