Inverse Equivalent Surface Current Method with Hierarchical Higher Order Basis Functions, Full Probe Correction and Multilevel Fast Multipole Acceleration (Invited Paper)
Inverse Equivalent Surface Current Method with Hierarchical Higher Order Basis Functions, Full Probe Correction and Multilevel Fast Multipole Acceleration (Invited Paper)
复制标题
DOI：
10.2528/pier10061604
复制
发表时间：
2010
影响因子：
6.7
通讯作者：
T. Eibert;Ismatullah;E. Kaliyaperumal;C. Schmidt
中科院分区：
文献类型：
--
作者：
T. Eibert;Ismatullah;E. Kaliyaperumal;C. Schmidt
研究了一种在适当选择的惠更斯面上使用等效电和/或磁面电流密度的逆等效面电流方法。所考虑的具有三角形面网格的模型与从表面积分方程的矩量法（MoM）解中已知的模型兼容。考虑了0.5阶和1.5阶的散度符合电流基函数，其中0.5阶函数是著名的拉奥 - 威尔顿 - 格利森基函数。通常从测量中获得的已知近场样本利用电流的辐射积分作为正向算子映射到未知的等效面电流密度上，其中测量探头的影响以类似于矩量法的加权积分形式表述。通过将多层快速多极子方法（MLFMM）适配到逆公式来加速正向算子的评估，其中MLFMM表示是通过仅使用测量探头天线的远场方向图进行完全探头校正的关键。由此产生的完全探头校正算法非常灵活且高效，发现计算速度主要取决于问题的MLFMM配置，而只要展开能够足够好地表示电流，就不太取决于特定的等效电流展开。针对各种问题展示了逆电流和远场方向图结果，其中考虑了从模拟以及实际测量中获得的近场样本。
An inverse equivalent surface current method working with equivalent electric and/or magnetic surface current densities on appropriately chosen Huygens surfaces is investigated. The considered model with triangular surface meshes is compatible with the models known from method of moments (MoM) solutions of surface integral equations. Divergence conforming current basis functions of order 0.5 and of order 1.5 are considered, where the order 0.5 functions are the well-known Rao-Wilton-Glisson basis functions. Known near-fleld samples typically obtained from measurements are mapped on the unknown equivalent surface current densities utilizing the radiation integrals of the currents as forward operator, where the measurement probe in∞uence is formulated in a MoM like weighting integral. The evaluation of the forward operator is accelerated by adaptation of the multilevel fast multipole method (MLFMM) to the inverse formulation, where the MLFMM representation is the key to full probe correction by employing only the far-fleld patterns of the measurement probe antennas. The resulting fully probe corrected algorithm is very ∞exible and e-cient, where it is found that the computation speed is mostly dependent on the MLFMM conflguration of the problem and not that much on the particular equivalent current expansion as long as the expansion is able to represent the currents su-ciently well. Inverse current and far-fleld pattern results are shown for a variety of problems, where near-fleld samples obtained from simulations as well as from realistic measurements are considered.