Partner: UPVD

Authors: Shab Gbémou, Julien Eynard, Stéphane Thil, Stéphane Grieu

Publicacion access here.

With the development of predictive management strategies for power distribution grids, reliable information on the expected photovoltaic power generation, which can be derived from forecasts of global horizontal irradiance (GHI), is needed. This work addresses the topic of multi-horizon forecasting of global horizontal irradiance using Gaussian process regression (GPR), and attempts to answer the following question: should time or past GHI observations be chosen as input? A comparison between time-based GPR models and observation-based GPR models is first made, along with a discussion on the best kernel to be chosen is each case; a comparison between horizon-specific and multi-horizon GPR models is then conducted. The forecasting results obtained are also compared to those given by the scaled persistence model. It is observed that, when seeking multi-horizon models, using a quasiperiodic kernel and time as input is favored, while the best horizon-specific model uses an automatic relevance determination rational quadratic kernel and past GHI observations as input. Ultimately, the choice depends on the complexity and computational constraints of the application at hand.