Automated Methods for Optimization of Aerospace Structures

The application of novel intelligent technologies in the search for optimal structures that meet both design and safety criteria has leveraged the automorphing of geometries based on a selection of design parameters. Given specific performance conditions, reserve factors may be predicted via machine learning surrogated models previously trained on datasets generated using design of experiments. We propose a new workflow that incorporates artificial intelligence to guide the automatic generation of structures with potential to comply with a certain objective function, e.g. the minimization of total weight. As a case study, this workflow is applied to a horizontal tail plane (HTP) multispar box with several decision parameters, showing a remarkable weight reduction as compared to the original structure. The implemented methodology accelerates the process of finding optimal configurations, is computationally sustainable and highly adaptable to other different scenarios.