Intermittence forecasting of the solar resource in Corsica
Abstract
Island electrical small-scale grids are sensitive to variations in power production. In case of important integration of solar power in the energy mix, the solar resource intermittency becomes a high risk to the grid’s stability. That is why a good knowledge of the variations is the first step to the massive optimized PV integration in the energy mix. This paper focuses on the forecasting of the solar resource variability. First, the solar resource variability characterization will be presented. It is based on a typological classification method that relies on variability scales which enable to distinguish the different dynamics. Afterwards, the predictability of variations is studied through forecasting models as a simple persistence, the k-Nearest Neighbors and Artificial Neural Networks (ANN). In this regard, time series of intervals classified according to their dynamics of variations have been generated and the forecasting performance, for different horizons and with different models were compared.
Keywords
Solar resource, Variations, Classification, Forecasting, K-nearest neighbors, Artificial neural networks