An Uncertainty Analysis Approach to Multidisciplinary Design Optimization

Abstract

Uncertainty management is an important research issue in multidisciplinary design optimization (MDO), especially at the conceptual design stage. The good uncertainty management can significantly improve the quality of product design. In this research, three types of uncertainties of MDO in conceptual design are analyzed, and a novel systematic approach is proposed to reduce these uncertainties. Uncertain variables are represented by probabilistic forms to reduce the variable uncertainties. A rough sets theory-based method is utilized to deduce the knowledge that can assist designers making configuration decisions. Furthermore, a re-sampled method is developed according to mean square error for reducing the uncertainties of the advanced kriging model. Finally, the validity and necessity of the proposed approach are demonstrated through the conceptual design of a bulk carrier.