The invention relates to the technical field of
data processing, in particular to an AI
new energy power prediction method, which comprises the steps of constructing an upstream candidate
station set; taking the target
station as a center, delimiting a
radius range in the upstream direction of the prevailing
wind direction, obtaining actually measured
wind speed data of adjacent stations in the range, and forming an upstream candidate
station set; for each upstream candidate station, calculating the spatial-
temporal correlation between the upstream candidate station and the target station, and selecting 1-3 optimal upstream target stations; constructing a dynamic
wind speed correction model; based on the screened optimal upstream target station, constructing a dynamic
wind speed correction model taking the NWP forecast wind speed as a reference and taking the actually measured wind speed of the upstream station as a core; and constructing and applying a power mapping model, and outputting final power prediction. According to the invention, through upstream and downstream linkage, wind speed correction is rapidly carried out on the downstream
electric field by using real-time change of the upstream wind speed, forecast wind speed data is adjusted in time, and a more accurate data basis is provided for power prediction.