kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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 Day-ahead and Intraday Electricity Market Prices with ARIMA, RNN-LSTM and TFT Models
KTH, School of Electrical Engineering and Computer Science (EECS).
KTH, School of Electrical Engineering and Computer Science (EECS).
2024 (English)Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Abstract [en]

Electricity price forecasting is a rapidly advancing field, driven by progress in datascience, the increasing necessity for accurate predictions due to global events, and the risingshare of variable renewable energy resources. This paper aims to forecast the hourlyelectricity prices for December 2023 in the Swedish bidding zone SE1 in the day-ahead andintraday electricity markets. First, a literature review of various forecasting methods isconducted. Then, three models - Auto-regressive Integrated Moving Average (ARIMA), LongShort-Term Memory (LSTM) recurrent neural network, and the Temporal Fusion Transformer(TFT) - are implemented. Performance of these models is evaluated based on their meanabsolute percentage error (MAPE), mean square error (MSE), and root mean square error(RMSE). In addition to price data, the models incorporate other external variables ascovariates, which are tested for each model and market, and included if found useful. TheLSTM model provides the lowest MAPE for day-ahead prices. The TFT achieves a lowerMSE and RMSE for day-ahead prices by better capturing price peaks. ARIMA performscomparably to the LSTM model for day-ahead forecasting. For intraday prices, the LSTMmodel delivers the lowest errors across all metrics. While the TFT captures the overall trend,it performs worse than the LSTM model, and ARIMA fails to produce useful intradaypredictions.

Abstract [sv]

Elprisprognosticering är ett snabbt framåtskridande område som drivs avframsteg inom data science, det ökande behovet av träffsäkra prognoser på grund av globalahändelser, samt den stigande andelen ombytliga förnybara energikällor. I denna rapportprognosticeras timpriset i december 2023 i Sveriges elområde SE1 på dagen före-ochintradagsmarknaderna. Först utförs en litteraturstudie av olika prognosticeringsmetoder.Därefter implementeras tre modeller - Auto-regressive Integrated Moving Average (ARIMA),Long Short-Term Memory (LSTM) recurrent neural network och Temporal FusionTransformer (TFT). Modellernas prestanda utvärderas avseende deras mean absolutepercentage error (MAPE), mean square error (MSE) och root mean square error (RMSE).Utöver prisdata använder modellerna andra externa variabler som covariates. Dessa testasför varje modell och marknad och inkluderas om de bedöms användbara. LSTM ger lägstMAPE för dagen före-priserna. TFT uppnår lägre MSE och RMSE för dagen före-prisernagenom att bättre fånga pristoppar. ARIMA ger jämförbar prestanda mot LSTM på dagen föreprognosticeringen. På intradagspriserna ger LSTM lägst fel enligt alla mått. Även om TFTfångar den övergripande trenden presterar den sämre än LSTM och ARIMA misslyckas attproducera användbara intradagsprognoser.

Place, publisher, year, edition, pages
2024. , p. 271-283
Series
TRITA-EECS-EX ; 2024:156
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
Identifiers
URN: urn:nbn:se:kth:diva-358456OAI: oai:DiVA.org:kth-358456DiVA, id: diva2:1928790
Supervisors
Examiners
Projects
Kandidatexamensarbete i Elektroteknik 2024Available from: 2025-01-17 Created: 2025-01-17

Open Access in DiVA

fulltext(121150 kB)71 downloads
File information
File name FULLTEXT01.pdfFile size 121150 kBChecksum SHA-512
82342f1408fe2aae929f55e76f2a176a8521cf94c0fe100464225724a9b74ddff6f61a0fb1cf957b5d6400be30877b9d0aec28080253ae7c1e09793536e10217
Type fulltextMimetype application/pdf

By organisation
School of Electrical Engineering and Computer Science (EECS)
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 71 downloads
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: 561 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • 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