Forecasting hourly electrical demand of European countries with transformer architectures

Forecasting hourly electrical demand of European countries with transformer architectures | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Forecasting hourly electrical demand of European countries with transformer architectures Eric Jahnke, Armin Ardone This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9671633/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Short-term forecasts of electrical demand are fundamental for grid operations by TSOs as well as a crucial input for analysis of players active on the electricity market. The more accurate the forecasts of future electricity demand are, the more precisely re-dispatch can be managed, the smaller the power imbalance that requires the use of balancing power, and the lower the risk of unplanned load shedding. Transformer architectures have been state-of-the-art in fields such as image recognition and natural language processing for years. One drawback of these particular deep learning models, however, is that they require a relatively large amount of training data to achieve good results. Hence, historical load data for many ENTSO-E bidding or control areas have only been available since 2015 in high resolution, meaning that for public holidays that occur only once a year, there are virtually no training samples available. Furthermore, social seasonality patterns were significantly disrupted during the Covid-19 pandemic, meaning that several months within the time series are partially corrupted and may distort the model training. Our approach demonstrates for three current Transformer architectures, that by selecting country-specific features, tuning hyperparameters, over-sampling and weighting during holiday weeks, it is possible to outperform forecasts from well-known Machine Learning or Deep Learning models as well as the official load forecasts from the ENTSO-E Transparency Platform, despite the limited number of training years. Other Economics Load forecasting Transformer architectures Feature selection mRMR Holiday effects Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9671633","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":637713830,"identity":"93955492-4b36-440a-9577-69d2df3c5c6d","order_by":0,"name":"Eric Jahnke","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA30lEQVRIiWNgGAWjYFACxgYGBh6GBCBmfNhgABJJIF4Ls2GDgQExWiAApIVNsoGBCC380ocbPzDI2OWZ8xw+Vjmj4A8DP3uOAV4tkn2JzRIMPMnFlr1taTc3AB0m2fMGvxaDM4xtQL8wJ244z2N28wFQi8ENArbYQ7TUA7XwfysEabEnpMWAB6zlcOKGsz1sjCCHGUgQ0CJxhrFZIoHneOKGM8eMJWcYGPNInHlWgFcLfw/7ww8fe6qBWpIffuz5IyfH3568Aa8WMEjsQbB5CCsHgx9EqhsFo2AUjIKRCQC4O0OCLIggRQAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0009-0006-8901-8128","institution":"Chair of Energy Economics, Institute for Industrial Production, Karlsruhe Institute of Technology","correspondingAuthor":true,"prefix":"","firstName":"Eric","middleName":"","lastName":"Jahnke","suffix":""},{"id":637713831,"identity":"ec6beee5-4fa4-4952-a70a-475cf5988108","order_by":1,"name":"Armin Ardone","email":"","orcid":"","institution":"Chair of Energy Economics, Institute for Industrial Production, Karlsruhe Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Armin","middleName":"","lastName":"Ardone","suffix":""}],"badges":[],"createdAt":"2026-05-10 16:35:21","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9671633/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9671633/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109760135,"identity":"ee9cbce4-d4c0-4508-b1a4-e9c0eb4f9a49","added_by":"auto","created_at":"2026-05-22 07:28:13","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2411903,"visible":true,"origin":"","legend":"","description":"","filename":"loadforecastingtransformersej2026.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9671633/v1_covered_9e975026-e8c7-46bf-982f-57e059c7aa85.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eForecasting hourly electrical demand of European countries with transformer architectures\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[{"identity":"c0b39b3d-f745-4c77-88dd-c5c94a120ca7","identifier":"10.13039/501100009318","name":"Helmholtz Association","awardNumber":"37.12.03","order_by":0}],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Karlsruhe Institute of Technology","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Load forecasting, Transformer architectures, Feature selection mRMR, Holiday effects","lastPublishedDoi":"10.21203/rs.3.rs-9671633/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9671633/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eShort-term forecasts of electrical demand are fundamental for grid operations by TSOs as well as a crucial input for analysis of players active on the electricity market. The more accurate the forecasts of future electricity demand are, the more precisely re-dispatch can be managed, the smaller the power imbalance that requires the use of balancing power, and the lower the risk of unplanned load shedding. Transformer architectures have been state-of-the-art in fields such as image recognition and natural language processing for years. One drawback of these particular deep learning models, however, is that they require a relatively large amount of training data to achieve good results. Hence, historical load data for many ENTSO-E bidding or control areas have only been available since 2015 in high resolution, meaning that for public holidays that occur only once a year, there are virtually no training samples available. Furthermore, social seasonality patterns were significantly disrupted during the Covid-19 pandemic, meaning that several months within the time series are partially corrupted and may distort the model training. Our approach demonstrates for three current Transformer architectures, that by selecting country-specific features, tuning hyperparameters, over-sampling and weighting during holiday weeks, it is possible to outperform forecasts from well-known Machine Learning or Deep Learning models as well as the official load forecasts from the ENTSO-E Transparency Platform, despite the limited number of training years.\u003c/p\u003e","manuscriptTitle":"Forecasting hourly electrical demand of European countries with transformer architectures","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-18 08:02:11","doi":"10.21203/rs.3.rs-9671633/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"949881f6-9fca-4538-ac7e-b947a9a27e55","owner":[],"postedDate":"May 18th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67873706,"name":"Other Economics"}],"tags":[],"updatedAt":"2026-05-18T08:02:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-18 08:02:11","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9671633","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9671633","identity":"rs-9671633","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}