Application of NNARX in Modeling a Solar Radiation Prediction

Solar energy defined as a radiant energy which emitted by the sun. Recent trend shows that solar energy had been an alternative power source to generate electricity using photovoltaic system. Solar radiation is the key measurement for this potential solar energy source but with the high cost to build and maintained the infrastructure had embark a new chapter with the implementation of prediction model. Many prediction models had been developed using various method including machine learning method and Artificial Neural Network (ANN) approach. Solar radiation prediction involves various non-linear parameter. That is why, a non-linear ANN method had been used. Non-linear Neural Network Autoregressive Model with Exogenous Input (NNARX) is a dynamic ANN method and had been widely applies to solve a non-linear dynamic time series prediction model. This paper will develop a NNARX to performed solar radiation prediction using both meteorological and measured data parameter for Malaysia. Performance analysis will be carried out using Mean Square Error (MSE) calculation based on actual and predicted data gain from the approach. Based on the result, it is shown that NNARX had given a significant prediction values on solar radiation with lowest MSE value of 0.0116.