Evaluation of the Deterioration of Ceramic Claddings by Application of Artificial Neural Networks

Abstract

The definition of models to evaluate the durability of buildings can be a challenging task due to the large set of variables, and their complexity, which affect the performance and durability of buildings. Recently, several methodologies have been developed to predict the service life of buildings and their components. The knowledge acquired through these models allowed for the adopting of more rational and sustainable solutions. This study proposes the application of artificial neural networks (ANNs) to estimate the service life of ceramic claddings of buildings in Brasília, Brazil. The variables that impact the service life of these claddings are identified through a stepwise regression analysis. Based on these variables, an ANN model was created to evaluate the degradation of ceramic claddings. The validity of the models proposed, a multiple linear regression (MLR) model and an ANN model, is discussed as well as the physical sense of the results obtained. According to similar studies and the empirical knowledge linked to the degradation of ceramic tile claddings, the models proposed lead to coherent results.