09:00
Session 5: Multiobjective Optimization
Chair: Bruno Sareni
09:00
20 mins
|
HIGHER-ORDER MULTIOBJECTIVE DESIGN OF MEMS
Paolo Di Barba, Antonio Savini, Slawomir Wiak
Abstract: An evolution strategy approximating a single Pareto-optimal solution in an objective space exhibiting more than two objectives, and a modification of a standard NSGA-II code controlling the spread-out of unfeasible solutions are proposed. The optimal design of a magnetic micro-switch is considered as the case study.
|
09:20
20 mins
|
A ROBUST MULTI-OBJECTIVE DESIGN OF ELECTROMAGNETIC DEVICES
Min Li, Rodrigo Silva, Armin Salimi, Frederico Guimaraes, David Lowther
Abstract: This paper discusses the practical issues of robust multi-objective design optimization problems such as uncertainty models and robustness metrics from both deterministic and statistical points-of-view. A robust design formulation is proposed for solving multi-objective design problems. In the novel method, the robustness of solutions, expressed as the maximum deviation from the nominal performance, is controlled by a constraint specified by the designer while the mean performance (approximated by the nominal value) is being optimized. This robust design formulation does not increase the number of objectives in the problem and is easy to incorporate into some of the existing multi-objective evolutionary algorithms (MOEA), such as the non-dominated sorting genetic algorithm (NSGA).
|
|
