The invention relates to a photovoltaic power prediction method based on solar radiation climate characteristic recognition, which comprises the following steps: 1) defining a weather type index SCF,and classifying and dividing weather state data corresponding to different times according to the weather type index SCF; 2) respectively selecting a slope radiation model for each type of weather state data to predict slope incident total radiation; 3) constructing a photovoltaic cell model to predict the direct-current power generation power of the photovoltaic array; and 4) constructing an inverter model, and calculating according to the photovoltaic array direct-current power generation power prediction value to obtain photovoltaic array alternating-current power generation power, therebyfinishing photovoltaic power prediction. Compared with the prior art, the method has the advantages of flexible application scene, weather type recognition, accuracy improvement, power prediction error reduction and the like.