Multi-wind-power-plant short-term power prediction method considering time evolution and space correlation

The invention discloses a multi-wind-power-plant short-term power prediction method considering time evolution and space correlation, and the method mainly comprises four modules: an input module carries out data collection and preprocessing, and objects are historical power and meteorological prediction data of a plurality of wind power plants in a target region; the time evolution mode tracking module extracts time sequence and multi-periodicity time evolution modes of historical wind power data through a gating circulation unit and a multi-kernel convolution layer; the space correlation mode attention module introduces a time-varying mode attention mechanism to endow correlation weights to different time evolution modes of multiple space variables; and finally, the output module outputs the day-ahead prediction scene of the multi-wind power plant power. According to the method, the time-space fusion multi-wind-power-plant short-term power prediction model with the deep learning capability is constructed, t