Solar Forecasting Performance using Artificial Intelligence Algorithms
Renewable energy sources, especially Solar Energy, are to play a larger role in Hybrid Generation in the upcoming and future energy and utility requirements. The importance of renewable energies has been growing at a fast pace, both because of the need to solve problems related to environmental issues and as a way of helping the increasingly difficult management of electricity grids. Solar power forecasting can solve number of the equality issues, in the concern of accurate forecasts of solar output from design and equipments. The techniques of Artificial Intelligence have already shown their effectiveness in tasks of high complexity, namely, Regression, Classification, and Forecasting. Also in the field of Renewable energies these tools can be extremely useful, in particular in the prediction of Solar Irradiance.we developed two algorithms in Python of prediction of Solar Irradiance based on two methods of Artificial Intelligence, which are the Artificial Neural Networks and the K-Nearest Neighbors Method. In the forecasting process, the models are trained with subsets of the one-year Solar Irradiance register in the city of Lisbon and then the next hour’s forecast is carried out. In order to understand the best method to perform predictions of solar irradiance among those studied, a comparative study between the models was carried out, taking into consideration the prediction errors and the simulation times of both models in the simulations made in different situations.