대체모델을 이용한 신뢰성기반 최적설계를 위한 효율적인 제한조건경계 샘플링 기법

Abstract

Reliability-based design optimization (RBDO) accompanies large computational cost due to reliability analysis. Surrogate models were introduced to overcome this computational drawback in RBDO. To ensure the accuracy of the reliability, the boundaries of constraints should be well approximated because the solutions of RBDO are commonly located near the boundaries of constraints. In an earlier research, constraint boundary sampling (CBS) was proposed to approximate the boundaries of constraints accurately by locating sample points on the boundaries of constraints. However, because of CBS located the sample points at all constraint boundaries, it can locate sample points unnecessarily far from RBDO solution. In this paper, efficient constraint boundary sampling (ECBS) is proposed to enhance the efficiency of CBS. ECBS uses the statistical information of kriging surrogate model to locate sample points on near the RBDO solution. ECBS is compared with conventional sampling techniques.