Multidisciplinary robust design optimization under parameter and model uncertainties

Abstract

ABSTRACTModel uncertainty, defined as the difference between the actual physical system and the computer model, is usually overlooked in robust optimization (RO). In addition, engineering systems are often comprised of several subsystems or disciplines. The coupling characteristic between different disciplines incurs the difficulty of solving multidisciplinary design optimization (MDO) problems. In this article, a new method is proposed for multidisciplinary robust design optimization (MRDO) incorporating parameter and model uncertainties. In this method, the parameter uncertainty is described by an interval model. The model uncertainty is quantified by the Bayesian approach, which adds a bias function into the computer model to offset the output discrepancy between the actual physical system and the computer model. Meanwhile, Gaussian process models of the computer model and the bias function are constructed. Finally, an MRDO framework considering parameter and model uncertainties is established. The performance of the proposed method is tested through two MDO problems.