A least squares support vector machine model for prediction of the next day solar insolation for effective use of PV systems

Abstract

Accurate prediction of daily solar insolation has been one of the most important issues of solar engineering. The amount of solar insolation on a given location is a vital data for photovoltaic plants. Systems efficiency is easily affected by the changes in solar radiation so, this study is aimed to develop a Least Squares Support Vector Machine (LS-SVM) based intelligent model to predict the next day’s solar insolation for taking measures. Daily temperature and insolation data measured by Turkish State Meteorological Service for three years (2000–2002) were used as training data and the values of 2003 used as testing data. Numbers of the days from 1st January, daily mean temperature, daily maximum temperature, sunshine duration and the solar insolation of the day before parameters have been used as inputs to predict the daily solar insolation. The simulations were carried out with SVM Toolbox of MATLAB software. As a conclusion the results show that LS-SVM is a good method in estimating the amount of solar insolation of a given location with 99.294% accuracy.