Ensemble forecast of photovoltaic power with online CRPS learning

Jean Thorey (1) , Christophe Chaussin (2) , Vivien Mallet (3)

1 CLIME2 - Coupling environmental data and simulation models for software integration
2 EDF R&D - EDF R&D
3 ANGE - Analyse du Génome

We provide probabilistic forecasts of photovoltaic (PV) production, for several PV plants located in France up to 6 days of lead time, with a 30-min timestep. First, we derive multiple forecasts from numerical weather predictions (ECMWF and Météo France), including ensemble forecasts. Second, our parameter-free online learning technique generates a weighted combination of the production forecasts for each PV plant. The weights are computed sequentially before each forecast using only past information. Our strategy is to minimize the Continuous Ranked Probability Score (CRPS). We show that our technique provides forecast improvements for both deterministic and probabilistic evaluation tools.

Mots clés: Machine learning; Probabilistic forecasting; Sequential aggregation; Ensemble forecasting; Photovoltaic power

Domaines: Mathématiques [math]

Jean Thorey, Christophe Chaussin, Vivien Mallet. Ensemble forecast of photovoltaic power with online CRPS learning. International Journal of Forecasting, 2018, 34 (4), pp.762-773. ⟨10.1016/j.ijforecast.2018.05.007⟩. ⟨hal-01909827v2⟩