Solar Radiation Prediction In Libya Using Neural Networks

In recent times, there has been a growing trend of utilizing neural networks in the field of prediction engineering, particularly in the domain of solar radiation predictions. This particular research study delves into the application of two types of neural networks, namely feedforward neural networks (BPNN) and radial basis function networks (RBFN), in the task of forecasting daily solar radiation levels across various cities in Libya. The models are trained using a comprehensive dataset from 25 cities across Libya, spanning a time period of 6 years. Key parameters such as latitude, longitude, and month are taken into consideration during the training process. Subsequently, the networks are put to the test using data that was not included in the training phase. The findings of this study reveal that the performance of the two networks differs significantly when subjected to testing. Notably, the radial basis function network emerges as the superior performer, exhibiting performance metrics of 93% compared to the backpropagation neural network's slightly lower performance of 93.15%.