Machine learning techniques for short term solar forecasting

Abstract

Paper presented to the 3rd Southern African Solar Energy Conference, South Africa, 11-13 May, 2015.

In this work, we propose a benchmarking of supervised machine learning techniques (neural networks, Gaussian processes and support vector machines) in order to forecast the Global Horizontal solar Irradiance (GHI). We also include in this benchmark a simple linear autoregressive (AR) model as well as a naive model based on persistence of the clear sky index. The models are calibrated and validated with data from Reunion Island (21.34°S ; 55.49°E). The main findings of this work are, that for hour ahead solar forecasting, the machine learning techniques slightly improve the performances exhibited by the linear AR and the persistence model. These nonlinear techniques start to outperform their simple counterparts for forecasting horizons greater than one hour.