Surrogate models for sheet metal stamping problem based on the combination of Proper Orthogonal Decomposition and Radial Basis Function.
Résumé
In this work, a combination of Proper Orthogonal Decomposition (POD) and Radial Basis Function (RBF) is proposed to represent the physics-base surrogate model. The method is tested on the deterministically parameterized process of deep drawing of sheet, which is a critical process in the problem of multi-parametric sheet metal forming. The basic idea is to consider design variables and/or uncertain parameters as additional dimensions for the solution of displacement. The order of the resultant high-fidelity model is reduced through the use of POD method. The POD basis, constructed from the snapshots of final displacements fields of FEM simulation, is used to determine the POD coefficients. The function coefficient is interpolated from these POD coefficients using RBF.