Reduction of springback by intelligent sampling-based LSSVR metamodel-based optimization

Abstract

This paper deals with the variable blank holder force in sheet metal forming in order to reduce springback effects after forming. A structural risk minimization principle-based metamodeling technique, least square support vector regression (LSSVR) method is applied to optimization. In order to improve the efficiency, an intelligent sampling strategy proposed by Wang et al. (Mater Des 30:1468–1479, 2009a) is integrated with the LSSVR. Therefore, the proposed strategies establish an adaptive metamodeling optimization system. The optimization procedure can be carried out automatically. To valid the flexibility of this system, the presented method is used to optimize the variable blank force parameters of the models from NUMISHEET’96 and torsion rail model. Compared with other popular metamodel-based optimization methods, the test results demonstrate the potential capability for nonlinear engineering problems.