The invention discloses a short-term wind speed prediction method applied to wind power generation, relates to a short-term wind speed prediction method, and aims to solve the problem that the prediction precision of an existing short-term wind speed prediction method of a wind power system still needs to be improved. The method comprises the steps of firstly acquiring a wind direction, a wind speed, a temperature, humidity, a pressure, a wind speed standard deviation and a wind direction standard deviation sample data; secondly preprocessing and normalizing the sample data, constructing an eigenvector and determining prediction sample training set and test set; and finally constructing a prediction function f(x) and performing short-term wind speed prediction. The method is suitable for wind speed prediction of a wind power field.