A Model-Based Parameter Estimation Technique for Wide-Band Interpolation of Periodic Moment Method Impedance Matrices With Application to Genetic Algorithm Optimization of Frequency Selective Surfaces

摘要

A model-based parameter estimation (MBPE) technique is introduced in this paper for efficiently interpolating periodic moment method (PMM) impedance matrices over a wide frequency band. In the model, only the Floquet harmonics that strongly affect the frequency band of interest are employed to approximate the matrix elements, while the contributions from all other higher-order harmonics are compactly represented by two additional terms. The derivation of the model is physics-based, and the objective is to find the coefficients of the terms in the model by utilizing the values of the impedance matrix elements calculated via PMM at only a few frequency points. The number and position of these fitting points can be pre-determined from the cutoff frequencies of the Floquet harmonics, which allows the MBPE interpolation process in this case to be completely automated. In other words, the number and position of the sampling points are only dictated by the periodicity of the frequency selective surface (FSS) structure and the frequency range of interest. Unlike many of the other scattering parameter-based techniques, the shape and the resonances in the response of the FSS do not have any impact on the construction of the interpolation model. This makes it particularly useful in genetic algorithm (GA) aided FSS design, since for a fixed periodicity and frequency range the MBPE interpolation is independent of the scattering response of candidate FSS designs. Several examples of the new PMM-MBPE approach are presented including one in which it is used to considerably speed up the GA-based design process for a reconfigurable FSS.