The behavior of the Photovoltaic Systems are based on solar radiation that focuses on the panels, thus the generation of energy depends on this resource. Planning of these systems makes that demanding energy can be interrupted due to variant radiation behavior. Very short term forecasting models can be useful to improve the available supplies. Present proposal allows a first approach for term shorter prediction of global solar irradiance. Linear and nonlinear models were compared to implement this forecasting. Results show that nonlinear models based on computational intelligence techniques provide better results with a simpler methodology to get the models.