A coupled model for energy production forecasting from photovoltaic farms

Authors: Marek Brabec, Emil Pelikán, Pavel Krč, Kryštof Eben, Marek Malý, Pavel Juruš
Publication year: 2011
Publication type: Abstrakt
Periodical: COST ES 1002 WIRE: Weather Intelligence for Renewable Energies
Bibliographic citation: BRABEC, Marek, PELIKÁN, Emil, KRČ, Pavel, EBEN, Kryštof, MALÝ, Marek, JURUŠ, Pavel. A coupled model for energy production forecasting from photovoltaic farms. In: COST ES 1002 WIRE: Weather Intelligence for Renewable Energies. Bern: Meteotest, 2011. Dostupné z: http://www.wire1002.ch/fileadmin/user_upload/Major_events/WS_Nice_2011/Spec._presentations/Brabec.pdf
ID: 376451

Abstract

In this paper, we will describe some of the problems that occur when a numerical weather prediction (NWP) model output is used to predict solar energy production (EP) from photovoltaic farms and present some solutions as well as directions for future developments. While it is obvious that the NWP output quality is of substantial importance in photovoltaic energy prediction, the calibration of NWP output to the EP looks deceptively simple on the surface. It might be surprising that, in fact, the calibration is far from being trivial and deserves a careful statistical treatment. Quite surprisingly, it leads to nonstandard statistical problems and models. Some of them, based on nonparametric regression and related techniques are presented in the paper. At the same time, we will stress some of the unusual features of the setup that are related to the so called errors-in-variables problem (they are induced by the fact that the NWP output cannot be perfect).