Fast Electromagnetic Algorithms Of Structures In Multilayered Media

With the fast development of computational electromagnetic techniques, the models for the multilayered media have been the efficient technique for simulating the complex electromagnetic environment. In this dissertation, the fast electromagnetic simulation of the microstrip integrated circuit and microstrip antenna arrays is investigated.In this dissertation, the Formulation-C spectral domain Green's functions for the analysis of the electromagnetic radiation and scattering by the perfectly conducting objects of arbitrary shape embedded in multilayered media are derived for the mixed potential integral equation (MPIE). The basis principle of the discrete complex image method (DCIM) is investigated and the equivalent denominator is first employed with a rigorous mathematic proof. By using real-coded genetic algorithm (RGA) and the Cauchy's integral theorem, the poles can be extracted so that the Green's function is computed accurately after the surface wave being properly dealt.Based on the efficient evaluation of Green's functions, the spatial method of moment (MoM) for the analysis of microwave integrated circuits and microstrip antenna arrays is investigated. In this dissertation, the modified fast Fourier transformation (FFT) technique is investigated in detail, including adaptive integral method (AIM), precorrected-FFT method, sparse matrix canonical grid method (SMCG), which has the advantages of excellent modeling capability as well as that of accelerating the matrix vector product by FFT. All these methods could reduce computation complexity to O(N log N) and the memory storage to O(N), where N is the number of unknowns. Several preconditioning techniques for these methods have been investigated in order to further reduce the solution time for iterative algorithms. Furthermore, the parallel AIM is investigated in this paper to further reduce the compution time, and the parallel preconditioning technique is employed to reduce the solution time and improve the parallel efficiency. The proposed method attains the fast electromagnetic simulation of the complicated and large-scale planar microstrip integrated circuit and microstrip antenna arrays.In this dissertation, the periodic structure in multilayered media--frequency selective surface (FSS) is also investigated. The spectral domain MoM is first investigated to analyze FSS. The spectral immittance approach is utilized to derive the spectral Green's functions with the (Rao-Wilton-Glisson) RWG basis functions and the Rooftop basis functions being utilized to discretize the surface currents, and for the later, the FFT technique can be used to accelerate the summation of the Floquet's modes so as to save the filling time for the impedance matrix. The spatial MoM is also investigated to analyze the FSS in this dissertation. The Ewald's method combined with Shank transformation method is utilized to accelerate the slowly converging series for the periodic Green's function. In the end, the genetic algorithm (GA) is investigated to optimize the element shape of the FSS by proposing a two-step fitness function to improve the optimization capability and speed up the convergence of the GA, and the recursive general minimal residual method (GMRESR) is utilized to further speed up the solution of the impedance matrix equation. Numerical results demonstrate the validity and efficiency of the presented method.