Data-knowledge fusion driven few-sample photovoltaic power integrated prediction method

The invention discloses a data-knowledge fusion driven few-sample photovoltaic power integrated prediction method. The method comprises the following steps: S1, generating a large amount of historical data related to a photovoltaic station to be predicted; s2, training the deep learning prediction model by using a large amount of generated historical data of the photovoltaic field station to be predicted, and establishing a first photovoltaic power generation prediction model; s3, training a prediction model; and S4, inputting a small number of samples of the target domain into the prediction model trained based on the source domain, and performing fine adjustment and migration of model parameters to obtain a second photovoltaic power generation prediction model. And S5, based on the photovoltaic power generation physical process, establishing a photovoltaic power generation prediction model to obtain a third photovoltaic power generation prediction model. And step S6, carrying out training and prediction of