The need to reduce the impact of traditional electricity generation necessitates an increase in the optimization of alternative systems that produce less environmental contamination. Renewables play a key role, with solar energy considered one of the most important energy supply sources. Solar power plants have to be perfectly designed to optimize electricity generation, and their placement must be as suitable as possible for the meteorological conditions. Clouds are the most mitigating factor in solar energy production and their study is decisive in locating the plant. Apart from the importance of studying clouds before building the solar plants, cloud detection is, equally decisive in adapting plant operation to cloud types during solar power plant operation. This adaptation benefits plant performance and allows electricity management to be integrated into the electricity grid. Nonetheless, the majority of cloud studies determine atmospheric parameters, which are sometimes not available. In this work, we have developed an automatic, fully-exportable cloud classification model, where Bayesian network classifiers were applied to satellite images so as to determine the presence of clouds, classifying the sky as cloudless or with high, medium and low cloud presence. There was an average success probability of 90% for all sky conditions. (C) 2016 Elsevier Ltd. All rights reserved.