순차적 크리깅 메타모델의 검증을 위한 민감도 수렴성 연구

Abstract

The metamodels have developed with a variety of design optimization techniques in structural engineering fields for the last decade because of efficiency, excellent prediction performance and easy interconnections into design frameworks. For constructing the metamodel, the sequential procedure such as design of experiments, metamodeling techniques and validation techniques is conducted. Because validation techniques measure the accuracy of the metamodel, the number of presampled points for the accurate kriging metamodel is decided by validation techniques. Because the interpolation model such as kriging metamodel based on computer experiments passes through responses at presampled points, the accuracy of the metamodel has been evaluated from comparison with real responses at same untried points with additional analyses. In this research, we study the convergency of sensitivity for efficient validation of the kriging metamodel for increasing of presampled points.