Prediction of solar radiation in Bejaia city using neurnal network (MLP)

In order to model the global solar radiation based on meteorological parameters for the Bejaia site, we established a database of more than 26,000 points obtained by recording every eight minutes of illumination and meteorological parameters (sunshine hours, ambient temperature, air pressure, relative humidity and rainfall). empirical models have been developed using several parameters and, recently, prognostic and prediction models based on artificial intelligence techniques such as neural networks. The daily averages were used to test NN models with 5 parameters and the relationship with the coefficient of the highest correlation was chosen. Two thirds were used to establish the model and one third for validation. We compared its performance with four models in the literature (Angstrom-Prescott, Bahel, Newland and Abdalla).