Estimation of Global Solar Radiation Using NNARX Neural Networks Based on the UV Index

Context: This work presents different models based on artificial neural networks, among them NNARX, for estimating global solar radiation from UV index measurements. The objective is to determine the efficiency of studied estimate in terms coefficient determination (R2), root-mean-square error (RMSE), and mean absolute (MAE). Methodology: It divided into four stages: i) conformation training dataset (in this case, it uses a set 213.019 data collected over five years city Pasto, Colombia, with Davis Vantage Pro 2.0 station); ii) pre-processing remove erroneous unusual data; iii) definition recurrent conventional networks according an analysis topologies, e.g. NNFIR NNARX; iv) evaluation estimation through metrics such as R2, RMSE, MAE. To validate model, new during last year was used, which not included training. Results: NNARX show best compared networks. NNARX221 model has RMSE 54,32 MAE 18,06 w/m2. Conclusions: are highly efficient at radiation, 0,9697 cases. most characterized by using two past times current instant, they feed their own estimated output. Furthermore, numerical results that contribution temperature relative humidity relevant improving radiation. These can be particularly important since only use measurements made sensors, less expensive than ones.