Short-term power prediction of hybrid wind power generation

Abstract Accurate wind power prediction is crucial for maintaining the stability of power systems during large-scale wind power grid integration. In this paper, the generalization ability and nonlinear prediction advantages of Long Short-Term Memory networks are combined with the data-smoothing characteristics of Grey Models for short-term wind power prediction. Experimental tests show that the improved method achieves an average R2 value of 82.47% in 24-h, 72-h, and 5-day wind power prediction tasks, effectively improving prediction accuracy and reducing prediction errors.