Intra-Day Direct Normal Irradiance Forecasting Under Clear Sky Conditions Using ANFIS

Rémi Chauvin (1) , Julien Nou (1) , Stéphane Thil (1) , Stéphane Grieu (1)

1 PROMES - Procédés, Matériaux et Energie Solaire

This paper deals with an intra-day Direct Normal Irradiance (DNI) forecasting methodology under clear sky conditions, for time horizon varying up to 5 h. The goal is to evaluate the variability and the needs of forecasting models under such conditions. It is a part of a project which aims to forecast the DNI under various sky conditions at short term horizon and high spatial resolution. A quick review of the clear sky DNI modeling is firstly performed. Then, the database selection and filtering process is described. Finally, the forecasting approaches are presented and compared to persistent models used as references. The most efficient model presented here is based on adaptive network-based fuzzy inference systems and the optimal configuration achieves very good forecasting results and validates the proposed methodology.

Mots clés: Adaptive Network-based Fuzzy Inference Systems; Direct Normal Irradiance; Clear Sky Models; Time Series Forecasting

Domaines: Automatique / Robotique; Traitement du signal et de l'image [eess.SP]

Rémi Chauvin, Julien Nou, Stéphane Thil, Stéphane Grieu. Intra-Day Direct Normal Irradiance Forecasting Under Clear Sky Conditions Using ANFIS. 19th IFAC World Congress, Aug 2014, Cape Town, South Africa. ⟨hal-01292455⟩