A Modified Adaptive Integral Method for Analysis of Large-scale Finite Periodic Array

Abstract

A fast algorithm based on AIM is proposed to analyze the scattering problem of the large-scale finite array. In this method, by filling zeros into the local transformation matrix, the near and far fields are isolated thoroughly to eliminate the near correction process. In the far part, a 5-level block-toeplitz matrix is employed to avoid saving the idle grids without adding artificial interfaces. In the near part, only one local cube is required to compute the local translation matrix and near impedance matrix, which can be shared by all elements. Furthermore, the block Jacobi preconditioning technique is applied to improve the convergence, and the principle of pattern multiplication is used to accelerate the calculation of the scattering pattern. Numerical results show that the proposed method can reduce not only the CPU time in filling and solving matrix but also the whole memory requirement dramatically for the large-scale finite array with large spacings.