ANN based prediction of daily global solar radiation for photovoltaics applications

Abstract

Measured value of minimum air temperature, maximum air temperature, average air temperature and solar radiation between 1 January 2012 to 31 April 2014 for Hamirpur city in Himachal Pradesh, India are used for prediction of daily global solar radiation (DGSR) with artificial neural network (ANN) technique. The prediction of DGSR are made with three combinations of input variables namely: (i) average air temperature, maximum air temperature and minimum air temperature, (ii) average air temperature and maximum air temperature, (iii) average air temperature. The results shows that ANN model with input variable as average air temperature and maximum air temperature predict DGSR with mean absolute percentage error of 5.35%. It can be used for predicting DGSR for sites where measured solar radiation is not available, proving useful for sizing of solar photovoltaic systems.