Open AccessImproving combined solar power forecasts using estimated ramp rates: data‐driven post‐processing approach
Author(s) -
Abuella Mohamed,
Chowdhury Badrul
Publication year - 2018
Publication title -
iet renewable power generation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.005
H-Index - 76
eISSN - 1752-1424
pISSN - 1752-1416
DOI - 10.1049/iet-rpg.2017.0447
Subject(s) - ensemble forecasting , wind power forecasting , solar power , computer science , consensus forecast , meteorology , forecast skill , solar resource , probabilistic forecasting , wind power , electric power system , power (physics) , environmental science , econometrics , solar energy , machine learning , engineering , mathematics , artificial intelligence , physics , quantum mechanics , probabilistic logic , electrical engineering
Several forecasting models are combined together to mitigate the uncertainty associated with the solar power generation resource and improve the power generation forecasts. The common ensemble approach in wind and solar power forecasting is the blending of meteorological data from several sources. In this study, the present and the past solar power forecasts, as well as the associated meteorological data, are incorporated into an ensemble learning tool. Since forecasts based on numerical weather prediction systems are more valuable in horizons longer than 6 h, the proposed approach includes the simple persistence model of hour‐ahead forecasts along with the different models of day‐ahead forecasts so that the combined forecasts become hour‐ahead solar power forecasts. In addition, the proposed approach combines the ramp rates of the forecasts to enhance the ensemble learning. Furthermore, the approach improves the ensemble learning by using two loss functions – the first function to minimise errors of the forecasts, and the second to minimise errors of the ramp rates of the forecasts. The performance of the combined forecasts is evaluated over the entire year and compared with other techniques.