Ultra-short-term photovoltaic power prediction based on multi-source data fusion with FreDF-iTransformer

Due to the increasing share of photovoltaic (PV) power generation in the power grid, the accuracy of PV power prediction can substantially affect the efficiency of power dispatch. In order to solve the problem of high-accuracy prediction of PV power, this paper proposes a PV power prediction algorithm based on multi-source data fusion. The method comprehensively considers multiple meteorological factors affecting PV power and enriches the calculation parameters of PV power through the combination of multi-source meteorological data and power data. In addition, the method improves the iTransformer model and further strengthens the model’s ability to extract time series information by introducing the FreDF module. Experimental results with other models on real datasets show that the proposed algorithm has higher prediction accuracy in PV power prediction.