Efficient Optimization of Microwave Structures Through Design Specifications Adaptation
Slawomir Koziel
School of Science and Engineering Reykjavik University, Iceland
Direct simulation-based design optimization of microwave structures involving numerous electromagnetic (EM) evaluations is typically impractical because of its high computational cost. Co-simulation [1] is only a partial solution because the circuit models with embedded EM components are still directly optimized. Efficient simulation-driven design can be realized using surrogate-based optimization (SBO) principle [2], where the optimization burden is shifted to a surrogate model, computationally cheap representation of the structure being optimized (fine model). The successful SBO approaches used in microwave area are space mapping (SM) [3], [4] and various forms of tuning [5], [6]. Both approaches can be efficient, unfortunately, their implementation is not straightforward. In particular, substantial modification of the optimized structure may be required (tuning), or additional mapping and more or less complicated interaction between auxiliary models is necessary (SM). Also, SM performance heavily depends on the selection of the surrogate model. Here, a novel technique is presented that exploits a circuit-based coarse model and the adaptive adjustment of the design specifications. Original design specifications are modified to take into account the difference between the fine and coarse model responses at the current design. The coarse model is then optimized with respect to the modified specifications to produce a new design that—assuming sufficient quality of the coarse model—gives a good prediction of the optimal fine model design with respect to the original specifications. The new method is extremely simple to implement, and, as demonstrated using several examples, it is able to yield a good design after one or two electromagnetic simulations.
IEEE Antennas and Propagation Society International Symposium (APSURSI)
2010
July
Due to copyright restrictions articles are not available for download. They are however indexed on this site and search results will therefore correctly indicate if such terms are present in the articles.
