Application of Artificial Neural Networks in the topological optimization of 2D structures

Resumo

Artificial Intelligence (AI) has increasingly permeated diverse fields of knowledge, enhancing processes across disciplines. Civil engineering is no exception, particularly in the area of structural engineering, where AI contributes to automating and optimizing design tasks. This research presents the development of an artificial neural network (ANN) implemented in MATLAB for the topological optimization of bidimensional structures subjected to in-plane loads. The study explores various structural geometries and evaluates different training algorithms to identify the best trade-off between computational efficiency and predictive accuracy. The ANN is trained using datasets generated from a classical topological optimization algorithm, where the input data describe the structural configuration and loading conditions, and the output data represent the optimized material layout. The proposed approach aims to accelerate the optimization process while maintaining the reliability of the results.