<p> This thesis contributes significantly towards the development of a robust algorithm for design sensitivity analysis and the optimization of microwave structures. Based on the frequency-domain finite-element method, the approach provides accurate sensitivity information using both 2-D and 3-D formulations. It also significantly accelerates the optimization process.</p> <p> The design sensitivity analysis method greatly influences the efficiency and accuracy of gradient-based optimization by providing the response gradient
(response Jacobians) for the whole range of parameter values. However, common commercial electromagnetic simulators provide only specific engineering responses, such as Z- or S-parameters. No sensitivity information is made available for further exploration of the design-parameter space. It is common to compute the design sensitivities from the response themselves using finite-difference or higher-order approximations at the response level. Consequently, for each design parameter of interest, at least one additional full-wave analysis is performed. However, when the number of design parameters becomes large, the
simulation time becomes prohibitive for electromagnetic design procedures.</p> <p> The self-adjoint sensitivity analysis (SASA) is so far the most efficient way to extract the sensitivity information for the network parameters with the finite-element method. As an improvement of the adjoint-variable method (AVM), it eliminates the additional (adjoint) system analyses. With one single full-wave analysis, the sensitivities with respect to all design parameters are computed. This significantly improves the efficiency of the sensitivity computations. Through our proposed method, the finite-difference frequency-domain self-adjoint sensitivity analysis (FDFD-SASA), the process is further improved by eliminating the need for exporting the system matrix, thus improving both compatibility and computation time. The only requirement for integrating the sensitivity solver with the commercial EM simulators is the ability to access the field solution at the user-defined grid points. The sensitivity information is obtained by simple manipulation of the field solution as a post-process and hence, it adds little or no overhead to the simulation time.</p> <p> We explore the feasibility of implementing our newly proposed method using field solutions from a frequency-domain commercial solver HFSS v 11. We confirm the accuracy of the FDFD-SASA for shape parameters of metallic structures. Both 2-D and 3-D examples are presented, where the reference results are provided through the traditional finite-difference approximations. Also, the efficiency of the FDFD-SASA is validated by a filter design example, exploiting
gradient-based optimization algorithm.</p> / Thesis / Master of Applied Science (MASc)
| Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/22364 |
| Date | 04 1900 |
| Creators | Hasib, MD Arshaduddin |
| Contributors | Nikolova, Natalia K., Electrical and Computer Engineering |
| Source Sets | McMaster University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
Page generated in 0.0015 seconds