Simulation-driven adjustment of geometry and/or material parameters is a necessary step in the design of contemporary antenna structures. Due to their topological complexity, other means, such as supervised parameter sweeping, does not usually lead to satisfactory results. On the other hand, rigorous numerical optimization is computationally expensive due to a high cost of underlying full-wave electromagnetic (EM) analyses, otherwise required to assess antenna performance in a reliable manner. Design closure normally requires a local search, often carried out by means of gradient-based procedures. In this work, accelerated trust-region gradient-search algorithm is proposed for expedited optimization of antenna structures. In our approach, finite differentiation conventionally used to estimate the antenna response Jacobian is replaced, for selected variables, by a rank-one Broyden updating formula. The selection of variables is governed by the alignment between the direction of the recent design relocation and the coordinate system axes. Operation and performance of the algorithm is demonstrated using a set of benchmark wideband antennas. Comprehensive numerical validation indicates significant computational savings of up to 70 percent that can be achieved without compromising the design quality in a significant manner.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1108/ec-01-2019-0023
- Category
- Publikacja w czasopiśmie
- Type
- artykuły w czasopismach
- Language
- angielski
- Publication year
- 2020
