Scholarlink

<p>The low computational efficiency of Kriging-assisted reliability-based design optimization (RBDO) has long been a challenging problem hindering its general application to engineering design problems. To address this gap, this study develops an adaptive Kriging-assisted single-loop approach (AK-SLA) to improve the efficiency of computation. To begin with, the Kriging model is updated by using the approximate minimum performance target point (MPTP) from the last iteration of Kriging-assisted SLA. Since the approximate MPTP directly obtained from the Kriging-assisted SLA tends to be inaccurate due to poor fitting of highly nonlinear constraints, a dimension reduction-based inverse reliability analysis method is employed to yield a more accurate MPTP on the basis of the established Kriging model in the previous iteration. To automatically select approximate MPTPs or more accurate MPTPs to update the Kriging models during the iteration process, an adaptive strategy in terms of the error of the Kriging model and combining the Kriging model with the Karush-Kuhn-Tucker optimality condition is constructed. Finally, the viability of the proposed AK-SLA is verified through two numerical examples and an engineering problem. The results demonstrate that the proposed AK-SLA exhibits satisfactorily high computational efficiency to solve complex RBDO problems with implicit constraints.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Adaptive Kriging-assisted single loop approach for reliability-based design optimization via inverse dimension reduction method

  • 期刊论文

  • 发表日期：

    2026年5月5日

查看全文
  • Nan Ye,
  • Shaoqiang Bai,
  • Dequan Zhang,
  • Meide Yang,
  • Fang Wang

摘要

The low computational efficiency of Kriging-assisted reliability-based design optimization (RBDO) has long been a challenging problem hindering its general application to engineering design problems. To address this gap, this study develops an adaptive Kriging-assisted single-loop approach (AK-SLA) to improve the efficiency of computation. To begin with, the Kriging model is updated by using the approximate minimum performance target point (MPTP) from the last iteration of Kriging-assisted SLA. Since the approximate MPTP directly obtained from the Kriging-assisted SLA tends to be inaccurate due to poor fitting of highly nonlinear constraints, a dimension reduction-based inverse reliability analysis method is employed to yield a more accurate MPTP on the basis of the established Kriging model in the previous iteration. To automatically select approximate MPTPs or more accurate MPTPs to update the Kriging models during the iteration process, an adaptive strategy in terms of the error of the Kriging model and combining the Kriging model with the Karush-Kuhn-Tucker optimality condition is constructed. Finally, the viability of the proposed AK-SLA is verified through two numerical examples and an engineering problem. The results demonstrate that the proposed AK-SLA exhibits satisfactorily high computational efficiency to solve complex RBDO problems with implicit constraints.