Prediction of daily global solar radiation on a horizontal surface in Uganda using artificial neural networks.
Butto, Peter Balyegisawa
MetadataShow full item record
In this research a five parameter artificial neural network model was developed to predict daily global solar radiation on a horizontal surface using input variables that include; sunshine hours, maximum temperature, minimum temperature, relative humidity and rainfall and global solar radiation as output. The data used was for Kampala, Uganda from 01st January 2011 to 31st December 2016. Results have shown good performance for the five parameter model with correlation between the measured and the predicted values of 0.95 having 120 neurons in the hidden layer. The transfer functions used in the hidden layer and output layer were triangular basis function (Tribas) and linear transfer function (Purelin) respectively. Levenberg-Marquardt training algorithm was used in the training process in the hidden layer. Mean bias error (MBE), root mean square error (RMSE) and mean absolute percentage error (MAPE) obtained were 0.07 MJ/m2/day, 1.49 MJ/m2/day and 7.55% respectively. This model predicted the daily global solar radiation with a relatively high accuracy as compared to other models.