Combined Forecasting Model Considering Wind Speed Attribute Reduction and Clustering

Abstract

Precise wind speed prediction is of great significance for large-scale wind power grid connection and system operation. The article puts forward a K-mediods clustering short-term wind speed prediction model based on the fast correlation reduction optimization. First the entropies of each wind speed attribute sequence and wind speed sequence are calculated. The fast correlation filtering algorithm is used to reduce the attribute dimensions and delete the redundant attributes. Then, the improved K-mediods is used to cluster the reduced wind speed data, and the optimal sequence of wind speed correlation attributes is obtained to ensure the information within the class to be accurate and comprehensive. The double-layer long and short time memory network is used to dig out the deep features and the short-term prediction. Finally, the practical wind speed of the wind field is predicted. Compared with the measured data, the precision and availability of the prediction model is proved. The results show that the method in the article has great advantages in the optimal method of the wind speed attribute data.