Using Artificial Intelligence to Analyze the Thermal Behavior of Building Roofs

This paper presents the application of an artificial neural network to model the thermal behavior of some roof coatings used in buildings. A set of test cells was built to evaluate these roof coatings. The cells were placed outdoors and several parameters were measured and collected for several weeks. The measured parameters included the temperature in different parts of the test cells. Additionally, the solar irradiance, the humidity, and the wind speed were measured and stored. We designed, built, and calibrated several heat flux transducers to measure the heat flux in each cell. Further, the reflectance and emissivity of the roof coatings were measured and used to create the model. The main contribution of this work is the modeling of an experimental system to evaluate the variability of the heat flux in building roofs using histograms. A statistical analysis based on computer simulations employing neural networks was performed to analyze those parameters that affect the heat flux in the roofs the most and the least. Finally, it was found that under specific conditions small increments in the reflectance of the coating can produce significant changes in the heat flux in the roof.