Estimation of the solar radiation using some meteorological data for the Mediterranean region with the artificial neural network

In order to estimate monthly average global solar radiation on a horizontal surface for selected 14 locations in Mediterranean region was used artificial neural network (ANN) model. The ANN architecture designed is a feed-forward back-propagation (FFBP) model with one-hidden layer containing 12 neurons with logaritmic sigmoid (logsig) as the transfer function and one output layer utilized a linear transfer function (purelin). The training algorithm used in ANN model was the Levenberg Marquand back propagation algorith (trainlm). The data between 1993-2010 based on seven meteorological parameters (monthly average air temperature, minimum soil surface temperature, soil temperature at depths of 5 cm, relative humidity, cloudiness, vapor pressure, sunshine duration) and five geographical parameters (station, month, altitude, latitude, longitude) were taken from Turkish State Meteorological Service (TSMS). These meteorological and geographical variables were used as input parameters to obtain monthly mean global solar radiation as output in ANN model. The datasets of 14 stations were split into two parts for training and for testing the data. Results obtained from ANN model were compared with measured meteorological values by using statistical methods. The correlation coefficient (R2), Root Mean Square Error (RMSE) Mean Absolute Percentage Error (MAPE), Mean Square Error (MSE), Root Mean Square Percentage Error (RMSPE), Mean Square Percentage Error (MSPE), Mean Absolute Bias Error (MABE), Mean Bias Error (MBE), Mean Percentage Error (MPE), Willmott's Index (WI) and t-test values were found to be 0.940 (%), 1.562 (MJ/m2) , 0.079 (%), 2.441(MJ/m2), 0.110(%), 0.012 (%), 1.072 (MJ/m2), -0.417 (MJ/m2), -0.035%, 0.983(MJ/m2), ve 6.211(%) , respectively. Results show good agreement between the estimated and measured values of global solar radiation. We suggest that the developed ANN model can be used to predict solar radiation another location and conditions.