Akademik Veri Yönetim Sistemi
2025 10th International Conference on Computer Science and Engineering (UBMK), İstanbul, Türkiye, 17 - 19 Eylül 2025, ss.196-201, (Tam Metin Bildiri)
In recent years, it is very important to make accurate energy production estimates for the sustainability of energy systems and effective resource management. In this study, monthly electricity production data obtained from solar, biomass, fossil gas and fossil oil resources of four European Union member countries namely Germany, Spain, Belgium and Austria were examined. The dataset has energy data of the countries from January 2019 to December 2024. The goal is to assess the effectiveness of several time series forecasting models in capturing diverse patterns across countries and energy types. Five forecasting models were employed in the estimation process: Long-Short Term Memory(LSTM), Gated Recurrent Unit (GRU), Convolutional Neural Network (CNN), Seasonal Autoregressive Integrated Moving Average (SARIMA) and Deep Neural Network (DNN). The findings show that deep learning models, especially DNN and GRU, generally perform better than classical statistical models such as SARIMA. The best performance was obtained by the DNN model on Belgium’s fossil oil dataset, with an RMSE of 0.0403 and an MAE of 0.0333, corresponding to a forecasting accuracy of approximately 96.7\%. Also GRU gave very successful results in biomass and fossil gas estimation. As a result of this study, it was seen that Artificial Intelligence-supported energy estimation systems will play an important role in correctly managing the energy supply that will occur in the future.