Infilling strategies play an important role in kriging based optimization, especially when computationally expensive simulations are involved. In order to improve the efficiency of constructing a high-precision kriging model, an improved expected improvement criterion (IEI) and a parallel infilling strategy are proposed based on the maximum expected improvement (EI) criterion. In the proposed parallel infilling strategy, new sample points are generated by employing IEI criterion coupled with EI criterion. During the improved sampling process, redundant and pseudo sample are deleted in order to avoid failure of constructing a kriging model. An improved weighted particle swarm optimization (WPSO) algorithm is proposed to improve optimization efficiency. The proposed parallel infilling strategy is applied to nonlinear function optimization and variable blank holder force (VBHF) optimization in a double-c stamped part. Based on the LHD and software DYNAFORM, kriging models between the VBHF and forming quality are constructed. Compared with the initial kriging models, the meat relative error of kriging models with the proposed parallel infilling strategy for the wrinkling and average thinning rate are reduced by 95% and 55%, respectively. The optimal VBHF is obtained by the WPSO. The results show that, cracking has been completely eliminated and wrinkling has been decreased, greatly improving the forming quality of the double-c stamped part.