A benchmarking of machine learning techniques for solar radiation forecasting in an insular context
| Autor: | Ted Soubdhan, Cyril Voyant, Mathieu David, Philippe Poggi, Philippe Lauret |
|---|---|
| Přispěvatelé: | Physique et Ingénierie Mathématique pour l'Énergie, l'environnemeNt et le bâtimenT (PIMENT), Université de La Réunion (UR), Sciences pour l'environnement (SPE), Université Pascal Paoli (UPP)-Centre National de la Recherche Scientifique (CNRS), Groupe de Recherche sur les Energies Renouvelables (GRER), Université des Antilles et de la Guyane (UAG), Université de Bordeaux (UB), Centre National de la Recherche Scientifique (CNRS)-Université Pascal Paoli (UPP) |
| Rok vydání: | 2015 |
| Předmět: |
machine learning techniques
Computer science 020209 energy Context (language use) 02 engineering and technology statistical models Solar irradiance Machine learning computer.software_genre Intraday solar forecasting 0202 electrical engineering electronic engineering information engineering General Materials Science ComputingMilieux_MISCELLANEOUS Artificial neural network Renewable Energy Sustainability and the Environment business.industry Statistical model Benchmarking 021001 nanoscience & nanotechnology Support vector machine Autoregressive model [SDU]Sciences of the Universe [physics] Benchmark (computing) Artificial intelligence [SDU.OTHER]Sciences of the Universe [physics]/Other 0210 nano-technology business computer |
| Zdroj: | Solar Energy Solar Energy, Elsevier, 2015, 112, pp.446-457. ⟨10.1016/j.solener.2014.12.014⟩ |
| ISSN: | 0038-092X 1471-1257 |
| DOI: | 10.1016/j.solener.2014.12.014 |
| Popis: | International audience; In this paper, we propose a benchmarking of supervised machine learning techniques (neural networks, Gaussian processes and support vector machines) in order to forecast the Global Horizontal solar Irradiance (GHI). We also include in this benchmark a simple linear autoregressive (AR) model as well as two naive models based on persistence of the GHI and persistence of the clear sky index (denoted herein scaled persistence model). The models are calibrated and validated with data from three French islands: Corsica (41.91°N; 8.73°E), Guadeloupe (16.26°N; 61.51°W) and Reunion (21.34°S ; 55.49°E). The main findings of this work are, that for hour ahead solar forecasting, the machine learning techniques slightly improve the performances exhibited by the linear AR and the scaled persistence model. However, the improvement appears to be more pronounced in case of unstable sky conditions. These nonlinear techniques start to outperform their simple counterparts for forecasting horizons greater than one hour. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect