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
Forecasting and Control Strategies for Industrial Flexibility in the Nordic Reserve Market
University of Gävle, Faculty of Engineering and Sustainable Development, Department of Building Engineering, Energy Systems and Sustainability Science.
2025 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

Rapid decarbonisation and the proliferation of wind and solar generation are transforming power systems. These changes create volatility that must be managed through flexible demand and responsive reserves. This thesis presents a comprehensive investigation into how machine‑learning models—particularly Random Forests—can support demand response by predicting system regulation states and translating those predictions into actionable control logic for an industrial load optimization. After reviewing more than adozen peer‑reviewed studies on energy forecasting, demand response and market design, the work constructs a rich dataset of meteorological and market variables aligned at hourly resolution. Cyclical encoding and interaction features capture the inherent periodicity of time and nonlinear relationships between weather and market signals. A Random Forest classifierand a regression model are trained on manual frequency restoration reserve(mFRR) activation data for 2023–2025. The classifier achieves an overall accuracy of 79 %, while the regression model (dedicated for the hourly loadprediction) explains 77 % of the variance in factory load. The predictions are then integrated into a scheduling algorithm that adjusts the load of a 30 MW industrial plant within the Swedish bidding zone SE1, respecting internal stock constraints for each process area. Compared with a baseline without cyclical and interaction features, the enhanced model improves the up‑regulation F1‑score from 0.65 to 0.74. The thesis concludes that interpretable machine‑learning, augmented by domain‑specific feature engineering,provides a robust and operationally viable foundation to maximize the bidingstrategy on mFRR market by integrating industrial consumers into balancing markets and facilitating the energy transition. We also quantify cross –modelalignment: using the model in August 2025, the schedule reduced load in~80% of hours classified as UP, stayed near-neutral on NONE hours, and achieved overall alignment of ~31% under a ±0.1 MW neutrality band(≈50% under ±0.5 MW).

Place, publisher, year, edition, pages
2025. , p. 36
Keywords [en]
Industrial flexibility Demand response Manual frequency restoration reserve (mFRR) Nordic balancing market Random Forest Load forecasting Predictive scheduling Cyclical feature encoding Interpretable machine learning Renewable energy integration
National Category
Energy Systems
Identifiers
URN: urn:nbn:se:hig:diva-49323OAI: oai:DiVA.org:hig-49323DiVA, id: diva2:2037472
Subject / course
Energy technology
Educational program
Energy systems – master’s programme (two years)
Presentation
2025-09-15, 09:14 (English)
Supervisors
Examiners
Available from: 2026-02-11 Created: 2026-02-11 Last updated: 2026-02-11Bibliographically approved

Open Access in DiVA

Babena, Thesis final version(2326 kB)33 downloads
File information
File name FULLTEXT01.pdfFile size 2326 kBChecksum SHA-512
bf610e69613bfad5c3fa61635915e55b192bbbb97832783a63e8e4c6ebe2599d77f932317be54ec95903a28a10316b157af5bb1e0a31b5453101d7801bfcb5b5
Type fulltextMimetype application/pdf

By organisation
Department of Building Engineering, Energy Systems and Sustainability Science
Energy Systems

Search outside of DiVA

GoogleGoogle Scholar
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 431 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