Hourly Solar Radiation Prediction Based on Nonlinear Autoregressive Exogenous (Narx) Neural Network

In this study, Nonlinear Autoregressive Exogenous (NARX) model was used to predict hourly solar radiation in Amman, Jordan. This model was constructed and tested using MATLAB software. The performance of NARX model was examined and compared with different training algorithms. Meteorological data for the years from 2004 to 2007 were used to train the Artificial Neural Network (ANN) while the data of the year 2008 were used to test it. The Marquardt–Levenberg learning algorithm with a minimum root mean squared error (RMSE) and maximum coefficient of determination (R) was found as the best in both training and validation period when applied in NARX model.