Ultra-short-term wind speed revision method and system based on wind speed daily fluctuation characteristics

The invention discloses an ultra-short-term wind speed revision method and system based on wind speed daily fluctuation characteristics, and the method comprises the steps: carrying out the wind speed daily segmentation, fluctuation characteristic extraction and classification of historical wind speed data through constructing a VSDA model, and providing wind speed daily data and wind speed fluctuation marks for a wind speed revision algorithm; correlation between wind speeds and wind speed fluctuation characteristics are highly summarized; and then, taking past wind speed daily data, corresponding fluctuation marks and numerical weather forecast wind speed prediction data as model input, and realizing numerical weather forecast wind speed prediction value revision in combination with a Bayesian optimized long and short term memory neural network algorithm. According to the revision method provided by the invention, the accuracy of the wind speed prediction value can be effectively improved, the reliability o