Akademik Veri Yönetim Sistemi
International Journal of Computational and Experimental Science and Engineering, cilt.10, sa.4, ss.1078-1084, 2024 (Scopus)
Microgrids (MGs) are decentralized energy systems that integrate Distributed Energy Resources (DERs), energy storage units, and advanced control mechanisms to ensure reliable power supply. Due to the intermittent nature of renewable energy sources, accurate load forecasting is crucial for the stable operation of MGs, particularly in both grid-tied and islanded modes. This study explores the performance of multiple forecasting techniques, including Linear Regression (LR), Regression Tree (RT), Support Vector Regression (SVR), Gaussian Process Regression (GPR), and Artificial Neural Networks (ANN), to predict MG load using historical load and meteorological data. The models were evaluated using comprehensive datasets that include calendar parameters and detailed weather metrics such as temperature, humidity, wind speed, and felt temperature. Performance was assessed through error metrics including Mean Absolute Percentage Error (MAPE), Mean Squared Error (MSE), and Root Mean Squared Error (RMSE). Among the tested models, the ANN model incorporating a full set of meteorological parameters achieved the best performance, with a MAPE value of 2.58%. These findings highlight the importance of integrating detailed meteorological data for load forecasting in MGs, providing a framework for more reliable energy planning and enhanced operational efficiency.