2026-05-20T15:57:42.783Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:1:10915 2025-01-05T21:22:31Z 1

Forecasting Solar Energy Substation Voltages Using Metaheuristic Optimized GRU 2024 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/1/10915 https://ieeexplore.ieee.org/abstract/document/10819080 L. Jovanovic N. Jovic A. Petrovic N. Budimirovic M. Zivkovic N. Bacanin As global energy demand grows, renewable sources offer a key alternative to fossil fuels. However, integrating these sources into power grids presents challenges, especially with supply unpredictability. This paper demonstrates how AI, specifically using a modified PSO algorithm to enhance GRU models, can predict solar energy output and improve grid balancing. Experiments on public datasets show that the optimized models achieve high accuracy, with the best models demonstrating a mean square error (MSE) of 0.024642 and an mean absolute percentage error (MAPE) of 0.000943 in hourly predictions, facilitating renewable energy integration. conferenceObject 1 4 10.1109/TELFOR63250.2024.10819080 2024 32nd Telecommunications Forum (TELFOR), IEEE, Belgrade, Serbia