Estimation of daily global solar radiation using temperature, relative humidity and seasons with ANN for Indian stations

Abstract

Global solar radiation (GSR) is an important parameter in the design of photovoltaic systems. An accurate knowledge of the GSR of a location is essential for the efficient design and utilization of photovoltaic systems. The main objective of the paper is to predict the daily GSR under clear sky conditions of any location on a horizontal surface, based on meteorological variables. The various parameters such as earth skin temperature, relative humidity (simply humidity), date and month of the year are used to estimate the daily GSR. In order to consider the effect of each meteorological variable on daily GSR prediction, six combinations of the meteorological parameters are utilized in training the artificial neural network (ANN). Two cases were considered to train the ANN. In one case three years data of Hyderabad and in other case three years data of three cities (total nine years data) namely Hyderabad, Delhi and Mumbai are used. In both the cases, 90 days of Trichy data is used for testing the network. Accuracy was tested with statistical indicators like root mean square error (RMSE), mean absolute percentage error (MAPE) and mean bias error (MBE). It is found that MAPE value is minimum when date, month, temperature and humidity are considered as input variables.