Control of robust design in multiobjective optimization under uncertainties

Tohid Erfani, Sergei V. Utyuzhnikov

Research output: Contribution to journalArticlepeer-review

31 Citations (Scopus)

Abstract

In design and optimization problems, a solution is called robust if it is stable enough with respect to perturbation of model input parameters. In engineering design optimization, the designermay prefer a use of robust solution to a more optimal one to set a stable system design. Although in literature there is a handful of methods for obtaining such solutions, they do not provide a designer with a direct and systematic control over a required robustness. In this paper, a new approach to robust design in multiobjective optimization is introduced, which is able to generate robust design with model uncertainties. In addition, it introduces an opportunity to control the extent of robustness by designer preferences. The presented method is different from its other counterparts. For keeping robust design feasible, it does not change any constraint. Conversely, only a special tunable objective function is constructed to incorporate the preferences of the designer related to the robustness. The effectiveness of the method is tested on well known engineering design problems.

Original languageEnglish
Pages (from-to)247-256
Number of pages10
JournalStructural and Multidisciplinary Optimization
Volume45
Issue number2
DOIs
Publication statusPublished - Feb 2012
Externally publishedYes

Keywords

  • Directed search domain
  • Fuzzy uncertainty
  • Multiobjective optimization
  • Robust design optimization

Fingerprint

Dive into the research topics of 'Control of robust design in multiobjective optimization under uncertainties'. Together they form a unique fingerprint.

Cite this