Pronóstico del recurso solar a corto plazo para Distritos industriales basado en redes neuronales artificiales

Abstract

Traditional and non-polluting energy is a relevant issue for humanity due to its exponential growth in recent years and the need for energy. Knowing that the largest source of energy available is solar energy, we must take advantage of this resource in order to create renewable energies that can replace non-renewable ones and, incidentally, reduce gas emissions into the atmosphere. For optimal use, photovoltaic and photothermal systems will be used, which entails the need for extensive knowledge about solar radiation that reaches the earth's surface, considering that these values change in a matter of hours, days or months depending on its latitude and the microclimates of each zone. For this research we will expose the training of an artificial neural network based on multilayers to predict a short-term horizon of solar radiation. The models that we consider vary according to the algorithm used in the training of an artificial neural network and this means that we have different power values in the simulation of a photovoltaic module.