Solar radiation prediction using metrological parameters

In this study, meteorological data, including maximum and minimum temperatures, wind speed, sunshine hours, degree of cloudiness, precipitation, pressure and humidity were collected in the sixth station of Mashhad, Isfahan, Ramsar, Zahedan, Urmia and Shiraz. The results showed that in all stations the maximum temperature and sunshine hours, in 5 stations degree of cloudiness and in 4 station pressure and wind speed component parameters affecting solar radiation. The most important influencing parameters are different at each station so that wind speed in 4 stations and sunshine hours in three stations are ranked first and second. Checking all parameters show that maximum temperature and the degree of cloudiness effective on solar radiation but is the least important parameters. After determining the effective meteorological parameters at each station, solar radiation was estimated using support vector machine (SVM) and three experimental methods Angstrom, Hargreaves and Abdullah. Hargreaves method estimate solar radiation less accurate than the two other experimental method. SVM estimate solar radiation in the test phase at Isfahan, Mashhad, Urmia, Ramsar, Shiraz and Zahedan stations by RMSE error 1.38, 1.28, 1.36, 1.51, 1.21 and 1.58 MJm-2d-1 and MEF error 3.59, 5.50, 4.18, 7.96, 3.26 and 5.17 percent. SVM estimate solar radiation with greater accuracy than empirical methods in all stations using artificial intelligence.