Wind power generation output power prediction method based on similarity theory

The invention discloses a wind power generation output power prediction method based on the similarity theory. The method comprises steps of: using an appointed climate type and an appointed weather type as the characteristic vectors of similar time segments, normalizing the climate type, and mapping the characteristic vectors of the weather type into numerical value; performing mode identification analysis; acquiring the validities of all data classifications of the characteristic vectors of the climate type, using the data classification with the lowest validity as a classification result to form a sample set of the similar time segments; constructing a factor vector sample set influencing the wind power generation system output power and acquiring the similarity of different time segments; establishing a machine learning model and training the machine learning model by using the output power of the similar time segments and the sample set of the similar time segments; and predicting the output power of the time segment by using the machine learning model. The wind power generation output power prediction method is simple and feasible, and accurate in prediction.