Forecasting solar power production by using satellite images

Stefanov, Dimitar (author), Demšar, Jure (author)

Increased integration of photo-voltaic capacities across the world has caused the electrical grid to become more difficult to maintain. In order to operate the grid in an efficient manner, its operators need to rely on accurate forecasts of the solar power generation. For this purpose, our work focused on predicting 2-h ahead solar power generation at 15 min intervals – a typical resolution requirement for solar power plants and system operation measurements. We made solar power generation forecasts for 233 different locations across Slovenia – more extensive research regarding the number of photo-voltaic locations than what can be found in the literature on this topic. We showed that the state-of-the-art deep learning architecture called Temporal Fusion Transformer outperforms well-established benchmarks in solar forecasting by significant margins across all metrics and training settings considered. Our investigation of feature importance also proved that the Temporal Fusion Transformer is capable of extracting a sufficient amount of information from satellite images meaning that such images can be used as a decent replacement for solar power data when the latter is not available.

• Language: English
• Keywords: solar forecasting, time series forecasting, deep learning, satellite images
• Work type: Article
• Typology: 1.01 - Original Scientific Article
• Organization: FRI - Faculty of Computer and Information Science
• Publication status: Published
• Publication version: Version of Record
• Year: 2025
• Number of pages: 13 str.
• Numbering: Vol. 300, Art.113823
• PID: 20.500.12556/RUL-175512
• UDC: 004.85:004.93:621.311.243
• ISSN on article: 0038-092X
• DOI: 10.1016/j.solener.2025.113823
• COBISS.SI-ID: 245205763
• Publication date in RUL: 30.10.2025

Record is a part of a journal

• Title: Solar energy
• Shortened title: Sol. energy
• Publisher: Association for Applied Solar Energy, Elsevier
• ISSN: 0038-092X
• COBISS.SI-ID: 5228039

Licences

• License: CC BY 4.0, Creative Commons Attribution 4.0 International
• Link: http://creativecommons.org/licenses/by/4.0/
• Description: This is the standard Creative Commons license that gives others maximum freedom to do what they want with the work as long as they credit the author.

Secondary language

• Language: Slovenian
• Keywords: napovedovanje proizvodnje sončne energije, napovedovanje časovnih vrst, globoko učenje, satelitske slike

Projects

• Funder: NSF - National Science Foundation
• Funding programme: Directorate for Geosciences
• Project number: 8200684
• Name: Diagnostic Analysis of Ecmwf/Fgge Data Fields in the South Pacific During January, 1979