Study of Sequential Radial Basis Function for Computation-intensive Design Optimization Problem

摘要

To enhance the efficiency of modern engineering optimization problems involving computation-intensive analysis models, metamodel-based optimizations become more and more attractive. The main contribution of this article is to develop a novel global optimization strategy using sequential radial basis function, notated as SRBF. In SRBF, significant sampling space method is proposed to successively increase samples in the region of interest and makes optimization process converge to the global optimum with high efficiency. SRBF is validated by using several benchmark numerical and engineering problems, and through comparison of other metamodel-based optimization method, SRBF shows satisfactory performance in both optimization efficiency and global convergence capability. Moreover, the robustness study demonstrates that SRBF possesses good robustness performance. Finally, the further work to enhance SRBF is discussed.