Solar Radiation Prediction Based on Machine Learning for Istanbul in Turkey

Abstract

The correct installation of solar energy systems is important for the energy efficiency of the system. The total solar radiation values reaching the system have an important role in determining the energy production potential of the solar energy system. In this study, statistical and machine learning methods used in solar radiation estimation are discussed. Forecasting methods are evaluated with the application on Istanbul region. The variability of the data collected for the Istanbul region is examined and the inappropriate data in the data are extracted. The data that are checked and approved are applied to the forecasting models and the models are compared and evaluated according to their error values. Models are evaluated according to variability values and error values over temporal horizons. Variability has an important role in determining the most appropriate forecasting model.