lnu.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Forecasting of electricity generation for unregulated hydropower plants with neural networks.
Linnaeus University, Faculty of Technology, Department of Mathematics.
2015 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

For energy companies, good estimates of their future power generation are both important for planning their resources to avoid shortages as well as for maximizing their profits in electricity trading. The aim of this thesis is to develop a model which can be employed to forecast the power generation of so-called Run-of-River hydropower plants, meaning their production cannot be regulated but solemnly depends on the current stream flow. 3-year data is provided for 10 hydropower plants by the company Bixia AB who is especially interested in improving their long-term (more than 24 hours ahead) forecasts. The chosen modeling approach is to employ articial neural networks with time lags of the power generation as input. Different designs are tested; in particular for long-term predictions direct and iterative forecasting is compared. In order to assess the neural networks, forecasts are made with previously unseen data and their accuracy is evaluated. Moreover, the predictions are compared to the forecasts of a SARIMA model that has been created earlier with the same data. The final neural network models forecast very well for the most part and outperform the previous SARIMA modeling approach. One of the obstacles is the long computational time for training the models, although this can be overcome by technological advancements. A main advantage of the neural network models is their adaptability to changes in the future, since they can be adjusted when new data becomes available.

Place, publisher, year, edition, pages
2015. , 128 p.
Keyword [en]
neural networks, machine learning, hydropower
National Category
Probability Theory and Statistics
Identifiers
URN: urn:nbn:se:lnu:diva-39790OAI: oai:DiVA.org:lnu-39790DiVA: diva2:787062
Educational program
Mathematics and Modelling, Master Programme, 120 credits
Supervisors
Examiners
Available from: 2015-02-10 Created: 2015-02-09 Last updated: 2015-02-10Bibliographically approved

Open Access in DiVA

No full text

By organisation
Department of Mathematics
Probability Theory and Statistics

Search outside of DiVA

GoogleGoogle Scholar

Total: 91 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • harvard1
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf