SVM-Based Topological Optimization of Tetrahedral Meshes

Abstract

Finite element analysis (FEA) has been widely used in various fields of industrial product analysis. During the whole process of FEA, mesh model generation plays a key role, which directly influences the accuracy and speed of FEA. In order to generate high quality mesh, a number of topology-based mesh optimization methods have been proposed and applied. However, they are all quite time consuming. In this paper, we propose a SVM-based approach to topological optimization of tetrahedral meshes, aiming to improve the efficiency of topological mesh optimization by using machine learning technique. First, the methodology of the SVM-based topological mesh optimization is put forward. Then the specific features for three kinds of flip operations for tetrahedral meshes are identified and the corresponding SVM models are further set up. Finally three SVM-based flip operations are implemented and the approach is verified and analyzed. The experiment result shows the SVM-based mesh optimization method can improve the mesh optimization efficiency without losing mesh quality.