hig.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard-cite-them-right
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • sv-SE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • de-DE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Hourly Hydropower Production Forecasting with Machine Learning: A Case Study in Linköping, Sweden
Linköping University Division of Energy Systems, Department of Management and Engineering, Linköping University, Linköping.
University of Gävle, Faculty of Engineering and Sustainable Development, Department of Building Engineering, Energy Systems and Sustainability Science, Energy Systems and Building Technology. Linköping University Division of Energy Systems, Department of Management and Engineering, Linköping University, Linköping.ORCID iD: 0000-0002-0604-3672
Linköping University Division of Energy Systems, Department of Management and Engineering, Linköping University, Linköping.
Linköping University Division of Energy Systems, Department of Management and Engineering, Linköping University, Linköping.
Show others and affiliations
2024 (English)In: Proceedings of the World Congress on New Technologies, 2024Conference paper, Published paper (Refereed)
Abstract [en]

Machine Learning (ML) is frequently utilized in prediction tasks; however, its applications in hydropower forecasting, particularly in forecasting hourly power production, has not been thoroughly investigated. In this paper, two Deep Learning (DL) models, namely an autoregressive neural network and Long Short-Term Memory, are compared to a seasonal autoregressive moving average (SARIMA) model to forecast the hourly power production at a hydropower station situated in Linköping, Sweden. Hyperparameter optimization algorithms are used to identify suitable DL models and algorithms for automatic model identification of SARIMA models are utilized. The three models are evaluated using a rolling origin strategy on a test dataset that consists of 10 months (January – October 2023) of hourly power production. The DL models provided similarly accurate forecasts as the SARIMA model according to mean squared error and mean absolute error. However, the DL models are poorly calibrated, resulting in lower coverage compared to the SARIMA model. Furthermore, the models are using a univariate time series (i.e., using historical power production to forecast future power production) and future studies need to explore additional variables that may be useful in providing a more accurate forecast.

Place, publisher, year, edition, pages
2024.
Keywords [en]
deep learning; forecasting; hydropower; Machine learning; power production; time series; uncertainty estimation
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:hig:diva-45830DOI: 10.11159/icert24.102Scopus ID: 2-s2.0-85205553408OAI: oai:DiVA.org:hig-45830DiVA, id: diva2:1905339
Conference
10th World Congress on New Technologies, NewTech 2024, Barcelona, 25-27 August 2024
Available from: 2024-10-14 Created: 2024-10-14 Last updated: 2025-10-02Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textScopus

Authority records

Milić, Vlatko

Search in DiVA

By author/editor
Milić, Vlatko
By organisation
Energy Systems and Building Technology
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar

doi
urn-nbn

Altmetric score

doi
urn-nbn
Total: 75 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard-cite-them-right
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • sv-SE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • de-DE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf