Improved Adaptive Cross Approximation Algorithm For Fast Analysis Of Electromagnetic Scattering / Radiation Problem

With the fast development of computational electromagnetic techniques, as a kind of integral equations method, Method of Moment(MoM) has been widely used in the analysis of the electromagnetic scattering and radiation problems.But it consumes more computation time and memory while the analysis object's electrical size increased. Fast multipole method(FMM) and multilevel fast multipole method(MLFMM) are often used to analyze the free space electromagnetic scattering. It can remarkably reduce numerical complexity of memory and CPU time of method of moment. However, in the MLFMM algorithm, the formulation, implementation, and occasionally performance depend on a prior knowledge of the Green's function. So it is hard to applied to solve some problems. For example, FMM algorithm would hard to modify Green's function when applied to layered media problems.For the shortcomings of MoM and FMM, this paper introduced a matrix-based compression algorithm-Adaptive cross approximation algorithm(ACA), and based on this study, an improved adaptive cross approximation algorithm is investigated in this paper.This paper describes the differences between the traditional adaptive cross approximation algorithm and other algorithms firstly, and expound its principles and processes in detail.On the basis, this paper emphasized on analysis the improved process of traditional adaptive cross approximation algorithm, so as to reduce the memory consumption and the time of iterative solution further. And laid the foundation for electromagnetic scattering and radiation of complex objects with electrically large size.This paper also introduce ACA into the study of the models of the multilayered media and the electromagnetic scattering and radiation of lossy media, which reflects the versatility of this algorithm.Finally, the parallel multilevel ACA is investigated in this paper, and used to analyze electromagnetic characteristics of frequency selective surfaces(FSS) with electrically large size.