A Data-driven Clear Sky Model for Direct Normal Irradiance

Solar energy is an ideal new energy source for power systems. In order to integrate solar energy into the power grid, an evaluation of the irradiance input to solar power systems is required in many applications. Clear-sky models describe the maximum input of solar power systems, which are particularly important for solar irradiance forecasting and numerical weather prediction. The existing clear sky models are empirical or semi-empirical models. In order to make use of the information in historical big data, a data-driven clear sky model for direct normal irradiance is proposed in this paper. Firstly, a clear sky detection algorithm is designed to collect clear-sky data from historical big data automatically. After that, the weighted k-Nearest Neighbors regression algorithm is used to calculate clear-sky direct normal irradiance by historical clear-sky data. The database of the National Renewable Energy Laboratory (NREL) is used for experiments to evaluate the performance of the proposed model, compared with other three clear sky models. The results of the experiments show that the proposed model is more accurate than other three clear sky models. (paper)