Photovoltaic Energy Forecasting via Artificial Neural Networks and Support Vector Machine Approaches
In recent decades, many Renewable Energy Systems particularly photovoltaics (PVs) are becoming important source for power generation and have been installed all over the world. However, frequently varying output of PV as a result of weather data including irradiation and temperature, soiling, cloud cover etc. make it an intermittent and unreliable source when connected to grid. For this reason, output forecasting plays an important role in the energy generation and implementation of solar power systems. Until recently, conventional and empirical solutions have been applied in traditional way to predict the solar energy. However, when the results of these approaches are examined, insufficient accuracy and other limitations are detected in the predictions obtained by traditional methods. To overcome these limitations, to deal with uncertainties and to handle the shortcomings of these traditional methods Artificial Intelligence based techniques come up with their strong and certain effectiveness. Artificial Neural Network (ANN) and Support Vector Machine (SVM) approaches are both artificial intelligence forecasting methods which maximize the accuracy of results of the real-world applications and gain the upper hand about execution speed and time. This study aims to compare the forecasting capabilities of ANN (using different learning functions for each ANN) and SVM approaches by providing the 2,5 year-historical data from the solar PV plant located in Simav, Kütahya as input. Predictions of the higher accuracy can be achieved by SVM using PV measurements with weather data including irradiance and temperature. Results showed that the SVM model performed better than ANN model.
Özge BALTACI – 1981 yılında Ankara’da doğdu. Hacettepe Üniversitesi Bilgisayar Mühendisliği Bölümü’nden 2004 yılında, Dokuz Eylül Üniversitesi Mekatronik Mühendisliği’nden 2010 yılında mezun olmuştur. Dokuz Eylül Üniversitesi Mekatronik Mühendisliği Bölümü’nde “Mekatronik Sistemlerde Büyük Veri Analizi” başlıklı doktora tezine devam etmektedir.