Data-driven day-ahead wind power prediction method

The invention discloses a data-driven day-ahead wind power prediction method. According to the method, the correlation relation of wind power and related characteristic factors thereof is considered, sensitivity indexes of wind power to historical wind power, wind speed, wind direction, air pressure, temperature and humidity key characteristics are constructed based on multiple correlation coefficient methods, and the correlation between the wind power and the related characteristics is quantified; sensitivity weighted similarity measurement is carried out, data similar to a to-be-predicted point is screened from a historical moment data set, and a historical similar data set is established; according to the meteorological prediction data of the to-be-predicted point, adaptively dividing a meteorological scene of a historical similar data set; and combining the distance weight and the scene division weight of the similarity measurement, carrying out weighted resampling, and averaging resampling results to rea