Ultra-Short-Term Prediction Method of Wind Power for Massive Wind Power Clusters Based on Feature Mining of Spatiotemporal Correlation

With the centralization of wind power development, power-prediction technology based on clusters has become an important means to reduce volatility power, so a large-scale method is proposed considering prediction stability. Firstly, fluctuating features farms are constructed by acquiring statistical further build divided model using fuzzy clustering algorithm. Then spatiotemporal data obtained attention network train in large scale. Finally, stability predictive performance analyzed comprehensive index evaluation system. The results show that RMSE lower than 0.079 at this paper experience 159 Nei Monggol Autonomous Region China and extreme error better 25% for total capacity farms, which indicates high accuracy.