A computationally efficient procedure for multi-objective optimization of antenna structures is presented. In our approach, a response surface approximation (RSA) model created from sampled coarse-discretization EM antenna simulations is utilized to yield an initial set of Pareto-optimal designs using a multi-objective evolutionary algorithm. The final Pareto front representation for the high-fidelity model is obtained using surrogate-based optimization techniques. A critical stage of the design process is an initial reduction of the design space aimed at estimating the region containing the Pareto set, which allows for low-cost construction of the RSA model even if the number of designable parameters is large. Illustration example is provided. An importance of the design space reduction is also demonstrated.
Authors
Additional information
- Category
- Aktywność konferencyjna
- Type
- publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
- Language
- angielski
- Publication year
- 2014
