Short-term wind power prediction method based on singular spectrum analysis

The invention discloses a short-term wind power prediction method based on singular spectrum analysis, and belongs to the technical field of wind power generation. Performing singular spectrum decomposition processing on the original wind power data to obtain a plurality of subsequences; performing convolution processing on the original wind power data and the plurality of subsequences to obtain a characteristic spectrum with wind power data local time correlation; processing the characteristic spectrum through a bidirectional gating unit network, and obtaining the global time correlation of the wind power data; and obtaining a wind power prediction result according to the global time correlation of the wind power data. According to the method, singular spectrum decomposition is utilized to process original wind power data to obtain a plurality of subsequences, feature extraction is carried out from multiple dimensions through convolution processing and bidirectional gating unit network processing, and the pr