Return to search

SIMULATOR INDEPENDENT EXACT ADJOINT SENSITIVITY ANALYSIS OF SELF-ADJOINT MICROWAVE STRUCTURES

<p>This thesis proposes a new analytical self-adjoint sensitivity analysis to calculate the Jacobian of the <em>S</em>-parameters for metallic shape parameters. This method is independent of the full-wave numerical analysis and the respective system matrix. The theory works for both volumetric and infinitesimally thin metallic shapes. It exploits the computational efficiency of the self-adjoint sensitivity analysis (SASA) approach where only one EM simulation suffices to obtain both the responses and their gradients in the designable parameter space.</p> <p>There are three major advantages to this development: (1) the Jacobian computation for metallic structures is completely analytical and there is no approximation involved in the sensitivity analysis of shape parameters; (2) the implementation is straightforward and in the form of a post-processing algorithm operating on the exported field solutions on the surface or around the edge of the metallic structure; and (3) it provides the possibility for exact sensitivity analysis with all electromagnetic high-frequency simulators whose system matrices are not available to export or are not differentiable with respect to shape parameters, e.g., simulators based on the FDTD method and the MoM.</p> <p>The method was verified in a number of examples using a commercial finite-element solver. The agreement between the results calculated with the proposed method and the reference self-adjoint sensitivity curves provided with the simulator are very promising.</p> <p>Suggestions for future work are provided.</p> / Master of Applied Science (MASc)

Identiferoai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/11103
Date10 1900
CreatorsDadash, Mohammad Sadegh
ContributorsNikolova, Natalia K., Bandler, John W., Mohamed H. Bakr, Ali Emadi, Electrical and Computer Engineering
Source SetsMcMaster University
Detected LanguageEnglish
Typethesis

Page generated in 0.0083 seconds