Very Short-Term Estimation of Global Horizontal Irradiance Using Data Mining Methods

Solar energy is one of the safe, clean and environment-friendly renewable energy sources. In solar energy systems, it is important to obtain the maximum solar irradiance during the day, predict the solar energy generation and increase the efficiency of solar systems. In this study, we focus on the veryshort term estimation of global horizontal irradiance utilizing knearest neighbor and Naive Bayes algorithms. In the estimation process, direct normal irradiance, diffuse horizontal irradiance, dry-bulb temperature and relative humidity parameters are used in the multi-tupled input structure. The k-nearest neighbor algorithm which employs direct normal irradiance and diffuse horizontal irradiance parameters in the 2-tupled input structure is observed as the most promising model with the lowest mean absolute percentage error.