Uncertainties in global radiation time series forecasting using machine learning: The multilayer perceptron case

Cyril Voyant (1, 2) , Gilles Notton (2) , Christophe Paoli (2) , Alexis Fouilloy (2) , Fabrice Motte (2) , Christophe Darras (2)

1 CHD Castellucio - CHD Castelluccio
2 SPE - Sciences pour l'environnement

As global solar radiation forecasting is a very important challenge, several methods are devoted to this goal with different levels of accuracy and confidence. In this study we propose to better understand how the uncertainty is propagated in the context of global radiation time series forecasting using machine learning. Indeed we propose to decompose the error considering four kinds of uncertainties: the error due to the measurement, the variability of time series, the machine learning uncertainty and the error related to the horizon. All these components of the error allow to determinate a global uncertainty generating prediction bands related to the prediction efficiency. We also have defined a reliability index which could be very interesting for the grid manager in order to estimate the validity of predictions. We have experimented this method on a multilayer perceptron which is a popular machine learning technique.We have shown that the global error and its components are essential to quantify in order to estimate the reliability of the modeloutputs.The described method has been successfully applied to four meteorological stations in Mediterranean area.

Mots clés: Stationarity; interval; Energy Prediction; Time Series Forecasting; Processing; Artificial Neural Networks

Domaines: Recherche d'information [cs.IR]; Génie des procédés

Cyril Voyant, Gilles Notton, Christophe Paoli, Alexis Fouilloy, Fabrice Motte, et al.. Uncertainties in global radiation time series forecasting using machine learning: The multilayer perceptron case. Energy, 2017, 125, pp.248-257. ⟨10.1016/j.energy.2017.02.098⟩. ⟨hal-01470994⟩