AURUM Mühendislik Sistemleri ve Mimarlık Dergisi

Yazarlar: Durhan GABRALI, Zafer ASLAN

Konular:Bilgisayar Bilimleri, Yapay Zeka

Anahtar Kelimeler:Artificial Neural Networks,Solar Energy,Solar Radiation Estimation,Short Term Forecast,Multiple Linear Regression

Özet: In recent years due to the depletion of fossil fuel resources and environmental concerns, the share of renewable energy resources in the current energy production show an increasing trend. Solar energy one of the most important renewable energy sources; is a clean, sustainable and environmentally friendly energy source. Therefore it is very important to know and estimate the solar radiation properties of the respective regions during the operation of existing solar energy systems or the installation of new systems. In this study meteorological data of 2016 (solar radiation, wind intensity, UV radiation index, surface 5 cm temperature, temperature 2 m, wind direction) belonging to Eskice region of Büyükçekmece District of Istanbul were used for a estimation solar radiation short and medium term. Multiple Linear Regression (MLR), and Artificial Neural Networks (ANN) which is one of the soft computing methods were used for modeling. All models were compared statistically. The results show that ANN models are more successful than MLR models in solar radiation estimation. Seasonally the most successful forecast for solar radiation was obtained in June. With the ANN model used in this stage for solar radiation R, nRMSE and MAPE values were calculated as 0.912, 48.5% and 28.21% respectively. Based on these data, it can be emphasized that solar energy potential can be estimated with sufficient reliability in the selected region and in June.

---