Rezervuar Hesaplama ile Enerji Tüketim Tahmini: Sivas İli Örneği

Energy consumption forecasting holds significant importance for the efficient use of energy resources and sustainable management. This study examines the application of Reservoir Computing (RC) to predict monthly energy consumption in Sivas, Turkey, over the period from 2015 to 2024. RC, a machine learning approach based on recurrent neural networks, processes temporal data using a fixed reservoir and trains only the output layer through linear regression. A simulated dataset tailored to Sivas includes energy consumption (kWh), temperature, and population data, collected over 120 months, with 80% designated for training and 20% for testing. The RC model, featuring 100 nodes, was implemented in Python and assessed using Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Results demonstrate higher accuracy compared to the traditional ARIMA method. This study highlights RC’s advantages in computational efficiency and precision, offering a practical solution for regions like Sivas, characterized by significant seasonal fluctuations. It is recommended that the model be further developed with real-world data in the future. This research showcases the potential of innovative approaches in energy planning.