Fully Automated Space Mapping Optimization of 3D Structures
J.W. Bandler, R.M. Biernacki and S.H. Chen
We present new results of fully automating the aggressive Space Mapping (SM) strategy
for electromagnetic (EM) optimization. The generic SM update loop and the model-specific
parameter extraction loop are automated using a two-level Datapipe architecture. We apply
the automated SM strategy to the optimization of waveguide transformers. We introduce a
multi-point parameter extraction procedure for sharpening the solution uniqueness and
improving the SM convergence. We present, for the first time, automated EM optimization
utilizing the commercial 3D structure simulator HFSS.
The Space Mapping (SM) concept combines the computational expediency of empirical engineering models and the acclaimed accuracy of electromagnetic (EM) simulators [1-3]. In our original work, the initial mapping is established by aligning the two models at a number of base points. Our recent aggressive SM strategy drastically reduces the upfront effort by targeting every EM simulation at optimizing the design and progressively refining the mapping using the Broyden update [3, 4].
To implement the SM strategy requires two nested loops: the iterative process of updating the mapping and targeting the next EM simulation; and the parameter extraction process of aligning the empirical and EM model responses. The difficulty of manually carrying out these steps might discourage some engineers from exploiting the benefits of the SM concept.
We present new results of fully automating the aggressive SM strategy, using a two-level Datapipe architecture . The outer level automates a generic aggressive SM loop including the Broyden update. The inner level implements parameter extraction for specific models, such as the Empipe interface to the EM simulator from Sonnet Software [5, 6].
We demonstrate the automated SM strategy on the optimization of both planar structures (a high-temperature superconducting filter) and 3D structures (waveguide transformers). We present, for the first time, automated SM optimization utilizing the commercial high-frequency structure simulator HFSS .
Parameter extraction is a crucial step in SM optimization. We investigate the impact of parameter extraction uniqueness on the convergence of the aggressive SM strategy. We introduce a multi-point parameter extraction approach to sharpening the solution uniqueness and improving the SM convergence.