Meta-inspired Bidirectional Memory Prediction Method for Wind Speed Reconstruction Clustering

The accurate prediction of wind speed is important for the integration of large-scale wind power and its safe operation. The complementary ensemble empirical mode decomposition with adaptive noise is employed to decompose the wind speed sequence into several modal components, while the fast correlation filtering is utilized to optimize these modal components and reduce the dimensionality to reconstruct the sample set. The Gaussian kernel distance is utilized to measure the sample spacing, and an initial value is selected to improve the k-medoids clustering, so as to improve both the clustering accuracy and stability of the high-dimensional sample space. The meta-heuristic optimization module is embedded into the bidirectional long short-term memory network to construct the meta-heuristic bidirectional memory network. The typical set training samples are input to optimize the built-in parameters, while the typical set test samples are input to optimize the structural parameters. Finally, the wind speed prediction value is generated. The wind field in Northeast China is taken as the research object, and the accuracy and generalization ability of the prediction model are verified.