Artificial neural networks for predicting global solar radiation in Al Ain City - UAE

The geographical location (latitude: 24° 16′ N and longitude: 55° 36′ E) of Al Ain city in the southwest of United Arab Emirates (UAE) favors the development and utilization of solar energy. This paper presents an artificial neural network (ANN) approach for predicting monthly global solar radiation (MGSR) on a horizontal surface in Al Ain. The ANN models are presented and implemented on 13-year measured meteorological data for Al Ain such as maximum temperature, mean wind speed, sunshine, and mean relative humidity between 1995 and 2007. The meteorological data between 1995 and 2004 are used for training the ANN and data between 2004 and 2007 are used for testing the predicted values. Multilayer perceptron (MLP) and radial basis function (RBF) neural networks are used for the modeling. Models for the MGSR were obtained using eleven combinations of data sets based on the above mentioned measured data for Al Ain city. Forecasting performance parameters such as root mean square error (RMSE), mean bias error (MBE), mean absolute percentage error (MAPE), and correlation coefficient (R²) are presented for the model. The values of RMSE, MBE, MAPE, and R² are found to be, respectively, 35%, 0.307%, 3.88%, and 92%. A comparison of estimated MGSR with regression models is carried out. The ANN model predicts better than other models. The estimated MGSR data are in reasonable agreement with the actual values. The results indicate the capability of the ANN technique over unseen data and its ability to produce accurate prediction models.