Extending structural optimization capabilities of FEA softs according to machine learning principles

Abstract

The purpose of this paper is to extend structural optimization capabilities of some FEA softs (NASTRAN/PATRAN, MARC-MENTAT) by coupling them with MATLAB codes in order to develop light weight design based on machine learning principles. We apply these ideas to some composite material with gyroid structure. The obtained results are interesting from both a theoretical and a practical point of view, shortening the design cycle for this type of materials.

Keywords

• structural optimization
• finite element simulations
• machine learning
• composite materials with gyroid structure
• stress analysis