Ultra-short Term Solar Irradiance Prediction Considering Satellite Cloud Images

With the growing adoption of solar photovoltaic (PV) power generation within the power system, ensuring the economical operation of the smart grid requires accurate prediction of PV power generation. Precisely estimating solar irradiance is critical for accurate prediction of PV power generation. Satellite cloud images provide information concerning both present and future cloud cover, which can be valuable in inferring solar irradiance. To precisely capture the impact of cloud cover on solar irradiance, this study proposed a model for ultra short term solar irradiance prediction considering satellite cloud images. By varying the inputs, three different models are constructed to investigate the effects of meteorological elements and satellite cloud images on solar irradiance prediction and applied to the prediction of GHI, DNI and DHI. Comparing the predicted and experimental results at different time scales, the results reveal that the model with processed satellite cloud information as input has more accurate precision for GHI and DNI prediction. The method proposed in the paper can be used for the ultra-short-term prediction of irradiance as well as for the prediction of photovoltaic power generation.