Confidence Interval-Based Robust Topology Optimization with Active Learning

Yueqi He

doi:10.1016/j.ress.2025.111904

Confidence Interval-Based Robust Topology Optimization with Active Learning

2025-11-04

0

PRE

AI

DOI：10.1016/j.ress.2025.111904

Original

Original request for help

Save

Abstract

En 中文

• A probabilistic confidence interval-based robust topology optimization (CI-RTO) method is proposed, which is non-intrusive and capable of accounting for complicated multi-source uncertainties. • The Kriging model is integrated in CI-RTO to accelerate the assessment of robustness, and a confidence interval active learning strategy is customized for the CI-based formulation to efficiently construct the Kriging model. • Sensitivity of the robust objective is derived analytically. In CI-RTO, the computational load of sensitivity analysis is only three times that of DTO, which ensures its high efficiency. • The proposed CI-RTO is applied to compliance minimization problems. Through four numerical examples, including both 2D and 3D, macro and micro design problems, the advantages and characteristics of CI-RTO are elaborated.

Confidence Interval-Based Robust Topology Optimization with Active Learning

Abstract

Journal

Organization