Design of ultra-thin composite deployable shell structures through machine learning
A data-driven computational framework is applied for the design of optimal ultra-thin Triangular Rollable and Collapsible (TRAC) carbon fiber booms. High-fidelity computational analyses of a large number of geometries are used to build a database. This database is then analyzed by machine learning to construct design charts that are shown to effectively guide the design of the ultra-thin deployable structure. The computational strategy discussed herein is general and can be applied to different problems in structural and materials design, with the potential of finding relevant designs within high-dimensional spaces.
© 2017 by Miguel A. Bessa and Sergio Pellegrino. Published by the International Association for Shell and Spatial Structures (IASS) with permission. The authors acknowledge financial support from the Northrop Grumman Corporation. Comments of an anonymous reviewer are gratefully acknowledged.
Published - MABessa_IASS17_MachineLearning_article_final.pdf