2026-05-04T19:52:50.524Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11575 2025-09-02T21:28:26Z 3

Photovoltaic Substation Voltage Forecasting Optimization Using Modified Metaheuristic and Gated Recurrent Unit Networks, Chapter in AIS Algorithms for Intelligent Systems: ICCI 2024: International Conference on Computational Intelligence, Springer 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11575 https://link.springer.com/chapter/10.1007/978-981-96-4539-8_1 V. Zdravkovic S. Ivanovic L. Jovanovic M. Zivkovic S. Tedic N. Bacanin As global energy demand continues to rise due to population growth, industrialization, and technological advancements, the limitations and environmental effects of conventional fossil fuel-based energy sources become more apparent. The transition to renewable energy is increasingly essential for long-term sustainability and reducing greenhouse gas emissions. AI-driven time series forecasting offers significant potential for improving the incorporation of renewable energy with existing power grids by enhancing the predictability of energy generation, which aids in balancing supply and demand more effectively. In this study, we investigate the use of optimization metaheuristics and lightweight gated recurrent unit (GRU) networks for substation voltage forecasting based on real-world data. A detailed comparative analysis is conducted on a publicly available dataset, where a modified version of the variable neighborhood search (VNS) optimizer is suggested. The study’s findings demonstrate that the best-performing models achieve high forecast accuracy, with an error rate of less than 1 V, highlighting the effectiveness of these methods in energy grid forecasting tasks. bookPart Springer, Singapore 1 13 10.1007/978-981-96-4539-8_1 AIS Algorithms for Intelligent Systems: ICCI 2024: Proceedings of International Conference on Computational Intelligence