Abstract
Neural networks were used to predict the energy generated by photovoltaic modules from climatic parameters in an Argentinean region. For this purpose, temperature, ambient relative humidity and wind speed data were collected over a year. Also, incident energy on the module plane, generated electric energy and module working temperature were measured. A very good estimation of the energy generated by modules was obtained from information about the geographical location and climatic parameters. According to our findings, even though direct and diffuse solar radiation data were unknown, neural networks may be used not only for an a priori evaluation of solar resource availability and electric energy generation, but also to define the optimum tilt angle of the photovoltaic installation.