Inteligência artificial aplicada ao monitoramento da saúde estrutural.

Abstract

In Brazilian society, civil construction is entirely linked to economic growth. As a result of this growth in the demand for civil construction and the limited availability of buildable areas available in large cities, building structures become increasingly slim and susceptible to excessive vibrations. The physical and geometric properties directly interfere in the rigidity of the structural system, which can reduce the load capacity of the structure and cause serious failures. A structural health monitoring system (SHM) emerges as a significant method to monitor the integrity of civil structures. Therefore, this research had as its main focus to advance the knowledge about the dynamic behavior of civil structures, seeking to evaluate a structural monitoring methodology based on Artificial Intelligence (AI) techniques, specifically, Artificial Neural Networks (ANNs) Backpropagation. To achieve this objective, in this research, Ansys Student software was used, which uses the Finite Element Method (FEM) to identify the dynamic properties of the structure, and MatLab to construct artificial neural networks. Furthermore, from the results obtained in the studied beam models, it could be observed that the use of Backpropagation Artificial Neural Networks to identify damaged elements was satisfactory.