Uluslararası Doğu Anadolu Fen Mühendislik ve Tasarım Dergisi

Yazarlar: Ahmet Oğuz GÖK, Ceyhun YILDIZ, Mustafa ŞEKKELİ

Konular:Mühendislik, Elektrik ve Elektronik

Anahtar Kelimeler:Solar Power Plant,Power Generation Forecast,Artificial Neural Network

Özet: Turkey has a large solar energy potential due to its geographical location and the installed power of Solar Power Pant (SPP) in the country is rapidly increasing. But the variability in SPP generation makes it a problem to operate these facilities in the electrical grid system. SPP generation forecasts are needed to solve this problem. In this study, a forecast system is proposed for SPP generations. Artificial Neural Networks (ANN) are used in the proposed forecast system. ANN is trained by using the Levenberg-Marquardt learning algorithm. In the training, verification and test processes of the ANN, the generation values of the SPP located in Kahramanmaras province and the cloudiness forecast data obtained from the Global Forecasting System (GFS) are used. Analyzes are performed by changing the structure and input values of ANN. As a result of the analyzes, it is found that the ANN that uses cloudiness forecast with the generation values as input is more successful than the ANN that uses only generation values.

---