Free-form feature classification for finite element meshing based on shape descriptors and machine learning

Finite element mesh generation (FE meshing) from three-dimensional (3D) computer-aided design (CAD) models is generally the most critical process in the finite element analysis pipeline. In the FE meshing, several manufacturers strictly prescribe the meshing patterns for specific classes of free-form features such as “boss” or “rib” features on CAD models and, thus, establish company-specific FE meshing rules of where and how many node points of elements should be placed over and inside a form feature, to ensure the analysis accuracy. However, these features are currently recognized and extracted manually by experienced engineers. Therefore, it is crucial for manufacturers to develop software where features such as bosses or ribs with complex free-form surfaces can be extracted from CAD models and categorized based on prescribed meshing rules, where an FE mesh for the feature region can be automatically generated in accordance with the rules in order to realize a high-quality and reliable finite element analysis (FEA) pipeline. To this end, an algorithm of the free-form feature classification for FE meshing of a triangular surface mesh generated from a CAD model is proposed in this paper, which utilizes 3D shape descriptors, Bag-of-Features, and machine learning techniques. By using the triangular mesh and machine learning, the classification algorithm enables a uniform and expandable feature. Moreover, it employs shape descriptors of a point feature histogram as a local surface descriptor and a thickness histogram as a global volumetric descriptor. A combination of both descriptors yielded more excellent classification performance accuracy (92%) and recalls (95%–98%) than a single descriptor. Additionally, the classification performance is almost not affected by the key point sampling density and visual word length. © 2020 CAD Solutions, LLC.