Deep Feature Combination Based Multi-Model Wind Power Prediction

As a new energy resource, wind power receives more and more attentions, and wind power prediction has become an important means to guarantee the normal operation of power grids. To get more accurate predicted results, a wind power prediction method based on deep feature combination and model fusion is proposed in this paper. Firstly, the feature selection method is applied to find important features. Secondly, the tree-based ensemble learning model XGBoost and LightGBM are adopted to construct high-dimensional combination features in parallel, and PCA is used to reduce the dimension of the high-dimensional combination features. Finally, the wind power is predicted by using the model fusion method. The wind power data of four different regions are used as the experimental data set. The experimental result shows that the accuracy of the proposed method is significantly improved compared with the single model methods and the common model integration methods.