ABSTRACT
The article presents an optimization of the load-bearing capacity of a steel-timber composite beam using artificial intelligence algorithms. The beam consisted of a cold-formed omega girder and a laminated veneer lumber slab. To achieve this goal, the authors developed three-dimensional parametric models of such composite beams using the finite element method (FEM). The parametric models served as the base for generating a comprehensive set of solutions, which were subsequently used to train a neural network. This neural network was designed to predict the behaviour of the composite beams and to optimize them. By training the neural network with this extensive dataset, the researchers aimed to create an efficient and accurate tool capable of reproducing the structural characteristics of composite beams.
