2026-05-19T18:35:25.841Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11916 2026-05-08T21:58:48Z 3

Solar Energy Production Forecasting with Long Short-Term Memory Networks Optimized by Modified Firefly Algorithm, Chapter in SST Studies in Smart Technologies: WCSC 2025: World Congress on Smart Computing, Springer 2026 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11916 https://link.springer.com/chapter/10.1007/978-981-95-0183-0_6 S. Malisic S. Andjelic S. Tedic B. Radomirovic L. Jovanovic N. Bacanin M. Zivkovic The demands for renewable energy are consistently raising in the recent years, since these sources offer significant upper hand over traditional fossil fuel sources. Despite clear advantages like sustainable production and eco-friendly nature, in practice, integrating these sources with the conventional power grids presents a unique set of challenges. The biggest problem by far is irregular supply of energy coming from renewable sources. A prospective solution for this obstacle is utilization of artificial intelligence (AI) methods for time series prediction, that can aid in keeping balance between production and consumption of the electricity originated from renewable sources. This study explores the application of long short-term memory (LSTM) networks to forecast the electricity produced by solar power plants. For achieving adequate execution, a modified version of the renowned firefly algorithm (FA) was suggested to adjust the model’s hyperparameters. The tuned models exhibit satisfactory outcomes, indicating suitability of the proposed approach for renewable energy forecasting challenge. bookPart Springer, Singapore 51 64 10.1007/978-981-95-0183-0_6 SST Studies in Smart Technologies: WCSC 2025: Proceedings of World Congress on Smart Computing