Reliability-based robust design optimization under hybrid random and interval uncertainties

Abstract

This paper utilizes the concept of robust design to conduct uncertainty-based design optimization under mixed random and interval uncertainties. Specifically, the reliability-based design optimization (RBDO) framework is used to handle random uncertainty, while the robust counterpart of RBDO is constructed using the minimax regret approach to handle interval uncertainty involved in the design. The constructed optimization model is a minimax optimization problem involving probabilistic constraints. In order to effectively solve the optimal robust design, the proposed solving process first decouples the minimax optimization problem, thereby transforming the solution of the original nested minimax problem into solving two simpler optimization tasks. Then, these two optimization tasks are iteratively solved in sequence by leveraging the single-loop approach, until the termination condition is met. The presented examples demonstrate the necessity of robust design and the feasibility of the proposed solution procedure.