Date
2023-03-14Subject
333 Economics of land and energy 004 Data processing and computer science TransferZeitreiheErneuerbare EnergienConvolutional Neural NetworkModellAuswahlAnpassungFotovoltaikWindenergieMetadata
Show full item record
Aufsatz
Model selection, adaptation, and combination for transfer learning in wind and photovoltaic power forecasts
Abstract
There is recent interest in using model hubs – a collection of pre-trained models – in computer vision tasks. To employ a model hub, we first select a source model and then adapt the model for the target to compensate for differences. There still needs to be more research on model selection and adaption for renewable power forecasts. In particular, none of the related work examines different model selection and adaptation strategies for neural network architectures. Also, none of the current studies investigates the influence of available training samples and considers seasonality in the evaluation. We close these gaps by conducting the first thorough experiment for model selection and adaptation for transfer learning in renewable power forecast, adopting recent developments from the field of computer vision on 667 wind and photovoltaic parks from six datasets. We simulate different amounts of training samples for each season to calculate informative forecast errors. We examine the marginal likelihood and forecast error for model selection for those amounts. Furthermore, we study four adaption strategies. As an extension of the current state of the art, we utilize a Bayesian linear regression for forecasting the response based on features extracted from a neural network. This approach outperforms the baseline with only seven days of training data and shows that fine-tuning is not beneficial with less than three months of data. We further show how combining multiple models through ensembles can significantly improve the model selection and adaptation approach such that we have a similar mean error with only 30 days of training data which is otherwise only possible with an entire year of training data. We achieve a mean error of 9.8 and 14 percent for the most realistic dataset for PV and wind with only seven days of training data.
Citation
In: Energy and AI Volume 14 (2023-03-14) eissn:2666-5468Sponsorship
Gefördert durch den Publikationsfonds der Universität KasselCitation
@article{doi:10.17170/kobra-202307268497,
author={Schreiber, Jens and Sick, Bernhard},
title={Model selection, adaptation, and combination for transfer learning in wind and photovoltaic power forecasts},
journal={Energy and AI},
year={2023}
}
0500 Oax 0501 Text $btxt$2rdacontent 0502 Computermedien $bc$2rdacarrier 1100 2023$n2023 1500 1/eng 2050 ##0##http://hdl.handle.net/123456789/14937 3000 Schreiber, Jens 3010 Sick, Bernhard 4000 Model selection, adaptation, and combination for transfer learning in wind and photovoltaic power forecasts / Schreiber, Jens 4030 4060 Online-Ressource 4085 ##0##=u http://nbn-resolving.de/http://hdl.handle.net/123456789/14937=x R 4204 \$dAufsatz 4170 5550 {{Transfer}} 5550 {{Zeitreihe}} 5550 {{Erneuerbare Energien}} 5550 {{Convolutional Neural Network}} 5550 {{Modell}} 5550 {{Auswahl}} 5550 {{Anpassung}} 5550 {{Fotovoltaik}} 5550 {{Windenergie}} 7136 ##0##http://hdl.handle.net/123456789/14937
2023-07-27T11:15:42Z 2023-07-27T11:15:42Z 2023-03-14 doi:10.17170/kobra-202307268497 http://hdl.handle.net/123456789/14937 Gefördert durch den Publikationsfonds der Universität Kassel eng Namensnennung 4.0 International http://creativecommons.org/licenses/by/4.0/ transfer learning time series renewable energies temporal convolutional neural network ensembles wind and photovoltaic power 333 004 Model selection, adaptation, and combination for transfer learning in wind and photovoltaic power forecasts Aufsatz There is recent interest in using model hubs – a collection of pre-trained models – in computer vision tasks. To employ a model hub, we first select a source model and then adapt the model for the target to compensate for differences. There still needs to be more research on model selection and adaption for renewable power forecasts. In particular, none of the related work examines different model selection and adaptation strategies for neural network architectures. Also, none of the current studies investigates the influence of available training samples and considers seasonality in the evaluation. We close these gaps by conducting the first thorough experiment for model selection and adaptation for transfer learning in renewable power forecast, adopting recent developments from the field of computer vision on 667 wind and photovoltaic parks from six datasets. We simulate different amounts of training samples for each season to calculate informative forecast errors. We examine the marginal likelihood and forecast error for model selection for those amounts. Furthermore, we study four adaption strategies. As an extension of the current state of the art, we utilize a Bayesian linear regression for forecasting the response based on features extracted from a neural network. This approach outperforms the baseline with only seven days of training data and shows that fine-tuning is not beneficial with less than three months of data. We further show how combining multiple models through ensembles can significantly improve the model selection and adaptation approach such that we have a similar mean error with only 30 days of training data which is otherwise only possible with an entire year of training data. We achieve a mean error of 9.8 and 14 percent for the most realistic dataset for PV and wind with only seven days of training data. open access Schreiber, Jens Sick, Bernhard doi:10.1016/j.egyai.2023.100249 Transfer Zeitreihe Erneuerbare Energien Convolutional Neural Network Modell Auswahl Anpassung Fotovoltaik Windenergie publishedVersion eissn:2666-5468 Energy and AI Volume 14 false 100249
The following license files are associated with this item: