Abnormal moment power restoration method and system based on multiple photovoltaic stations

The invention belongs to the technical field of electric power energy, and the method comprises the steps: collecting the data of similar power stations at a to-be-repaired moment, and obtaining the geographic coordinates of each photovoltaic power station; photovoltaic power stations under similar weather conditions are determined, and similar power stations are determined by adopting a fuzzy clustering algorithm with improved similarity; and establishing an offline data correction model, taking the clustering result data of the similar power stations as training samples, inputting the training samples into the data correction model for correction, and obtaining an actually measured power data correction result at the to-be-repaired moment. According to the invention, through a similarity identification method which comprehensively considers the distance correlation degree and the curve shape correlation degree, the accuracy of similar power station identification is improved; meanwhile, by constructing the