基于国产众核超级计算机的6×105核并行矩量法

为实现电磁计算的安全可靠和自主可控,该文基于“天河二号”国产众核超级计算机平台,开展大规模并行矩量法(MoM)的开发工作。为减轻大规模并行计算时计算机集群的通信压力以及加速矩量法积分方程求解,通过分析矩量法电场积分方程离散生成的矩阵具有对角占优特性,提出一种新型LU分解算法,即对角块矩阵选主元LU分解(BDPLU)算法,该算法减少了panel列分解的计算量,更重要的是,完全消除了选主元过程的MPI通信开销。利用BDPLU算法,并行矩量法突破了6×105 CPU核并行规模,这是目前在国产超级计算平台上实现的最大规模的并行矩量法计算,其矩阵求解并行效率可达51.95%。数值结果表明,并行矩量法可准确高效地在国产超级计算平台上解决大规模电磁问题。     

PC群集系统中并行矩量法研究

目前复杂环境的EMC分析,计算量过大,耗时太长,使分析与优化遇到实质性的困难.本文研究MPI结合矩量法的并行技术,测试了在PC群集系统中并行矩量法的性能,并以此完成大型计算.与大型机和工作站相比,PC群集系统中的并行矩量法可以用较低的硬件成本有效的实现复杂环境电磁特性分析.     

Étude des structures planaires en très hautes fréquences par une nouvelle formulation de la méthode des moments dans le domaine spatial

An efficient analytical integration technique for computation of spatial method of moments (MoM) integrals in conjunction with numerical matched loads is presented. The current distribution on the device is solved by using the well-known Galerkin’s MoM procedure applied to mixed potential integral equation in the spatial domain. The scattering parameters are determined by considering infinite lines at each port where only the fundamental mode is assumed to propagate. The contribution of this work is the development of an integration technique for the computation of spatial domain integrals, that is fast and rigorous. This technique is based on a Taylor series expansion of the integrands involving only polynomial functions. The use of polynomial forms in the integrals leads an immediate analytical integration, and the computation time will be considerably reduced.     

A fast MoM calculation technique using sinusoidal basis and testing functions for a wire on a dielectric substrate and its application to meander loop and grid array antennas

The input impedance matrix element of the method of moments (MoM) for an arbitrarily shaped wire antenna printed on a dielectric material Z/sub m,n/ is formulated to be composed of three terms Z/sup /spl psi/s//sub m,n/, Z/sup /spl psi///sub m,n/, and /spl Delta/Z/sub m,n/ involving single-, double-, and triple-integral calculations, respectively. The MoM based on the Z/sub m,n/ formulated in this paper (new MoM) is applied to two antennas-a meander loop antenna and a grid array antenna-as well as a simple loop used as a reference antenna. The computation time to obtain the current distribution of each antenna by the new MoM technique is compared with the time required for the conventional MoM, which has an impedance matrix element composed of four terms, all involving triple-integral calculations. It is revealed that the new MoM drastically reduces the computation time: for example, by a factor of 937 for the grid array antenna. In addition, the radiation characteristics of these two antennas are discussed. It is found that a reduced-size meander loop (62% smaller than the simple loop reference) has a radiation pattern similar to the simple loop reference. It is also found that the grid array has an axial beam radiation pattern without side lobes in the principal planes.     

New method for the efficient summation of double infinite series arising from the spectral domain analysis of frequency selective surfaces

When the method of moments (MoM) in the spectral domain is applied to the analysis of frequency selective surfaces, the entries of the MoM matrix are slowly convergent double infinite series. In this paper, a two-step acceleration technique is developed which makes it possible the fast and accurate computation of these double series in the particular case where subsectional rooftops are used as basis functions. The technique is based on a combination of the use of Kummer's transformation, the use of Poisson's transformation, and the determination of judicious Chebyshev polynomial interpolations of some of the spectral discrete functions involved in the infinite series. The results obtained show that when all the double series of the MoM matrix are to be computed with an accuracy of three significant figures, the new acceleration technique turns out to be about one thousand times faster than brute-force computation, and a few times faster than the acceleration technique based on fast Fourier transform.     

磁场积分方程矩量法中基函数的立体角修正

本文通过考察电磁散射问题矩量法求解中电场积分方程和磁场积分方程的公式 ,分析了在使用三角屋顶基函数情况下传统的磁场积分方程在计算带有棱角的电小尺寸金属物体雷达截面时存在的不足 ,提出了一种基函数的立体角修正技术 ,从而达到了减小计算误差的目的。计算结果表明了算法的有效性。     

高阶MoM及其与PO的混合法计算电磁散射问题

本文提出了一种新的以高阶矩量法(MoM)与物理光学法相结合的混合法(MoM-PO).该方法采用曲面参数化的离散方法,保证了建模的精确性.计算过程中将散射表面灵活划分为MoM区和PO区,在各自区域可以灵活确定离散单元的大小和密度.MoM区域的高阶矩量法,采用基于Lagrange插值的高阶矢量基函数,结合点匹配技术,比传统的高阶法简单,易于实现.计算结果表明,本文的高阶矩量法及其与物理光学法结合的混合方法能准确有效的计算目标的电磁散射特性.     

基于超级计算机的矩量法性能分析与优化

复杂目标的精确电磁特性分析往往需要巨大的存储和极长的计算时间。针对这一问题,结合国内发展迅速的超级计算机系统,研究了具有精确高效仿真能力的高性能电磁算法——高阶矩量法。提出了单元预选法来消除矩阵并行填充过程中的无效计算,加速矩阵填充过程。提出了一种具有更少的通信次数和通信量的新型并行LU分解算法,加速矩阵方程求解过程。数值测试表明提出的矩阵并行填充算法和矩阵方程并行求解算法在超级计算机平台上都能获得较高的并行性能,大幅提高了矩量法的仿真能力。     

Higher order hybrid method of moments-physical optics modeling technique for radiation and scattering from large perfectly conducting surfaces

An efficient and accurate higher order, large-domain hybrid computational technique based on the method of moments (MoM) and physical optics (PO) is proposed for analysis of large antennas and scatterers composed of perfectly conducting surfaces of arbitrary shapes. The technique utilizes large generalized curvilinear quadrilaterals of arbitrary geometrical orders in both the MoM and PO regions. It employs higher order divergence-conforming hierarchical polynomial basis functions in the context of the Galerkin method in the MoM region and higher order divergence-conforming interpolatory Chebyshev-type polynomial basis functions in conjunction with a point-matching method in the PO region. The results obtained by the higher order MoM-PO are validated against the results of the full MoM analysis in three characteristic realistic examples. The truly higher order and large-domain nature of the technique in both MoM and PO regions enables a very substantial reduction in the number of unknowns and increase in accuracy and efficiency when compared to the low-order, small-domain MoM-PO solutions. The PO part of the proposed technique, on the other hand, allows for a dramatic reduction in the computation time and memory with respect to the pure MoM higher order technique, which greatly extends the practicality of the higher order MoM with a smooth transition between low- and high-frequency applications.     

MoM-NPO混合算法分析天线-载体系统

安装在电大尺寸平台上的天线,载体的影响不容忽视。采用同样基于电流展开的矩量物理光学混合算法,并通过NURBS物理光学法进一步降低复杂结构的建模和计算,将载体对天线的影响用物理光学电流合并到包括天线的在内较小的矩量法区,使矩量法解决天线-载体系统成为可能。通过矩量法和混合算法分析载体对天线性能的影响证明,载体对天线的影响是显著的,混合算法能够在保证精度的前提下提高运算速度,为解决电大尺寸载体上天线的特性提供了良好的途径。     

Estimation of spurious radiation from microstrip etches usingclosed-form Green's functions

The problem of spurious radiation from electronic packages is considered by investigating the power radiated from microstrip etches that are excited by arbitrarily located current sources and terminated by complex loads at both ends. The first step in the procedure is to compute the current distribution on the microstrip line by using the method of moments (MoM). Two contributions of this paper are: (i) employing the recently derived closed-form Green's functions in the spatial domain, which permits an efficient computation of the elements of the MoM matrix; and (ii) incorporating complex load terminations in a convenient manner with virtually no increase in the computation time. The computed current distribution is used to calculate the spurious radiated power, and the result is compared with that derived by using an approximate, transmission line analysis     

一种快速计算雷达散射截面空域特性的方法

将最佳一致有理逼近理论和矩量法(MoM)相结合,提出了一种新的二维外推技术并对任意形状复杂目标雷达散射截面的空域二维特性进行了快速计算.与渐进波形估计(AWE)技术相比,Maehly逼近的优势在于不需要计算阻抗矩阵元素的高阶导数,能在更宽的范围内精确逼近MoM计算结果,并且很容易和MoM计算机代码相结合.与MoM逐点计...     

Efficient evaluation of the [Z] matrix with method of moment in grounding analysis by using adaptive spatial sampling approach

The design of grounding system often relies on the performance simulations. The evaluation of the current distribution of grounding system for many frequencies by using the method of moments (MoM) may take a long time since the impedance matrix must be recomputed at each new frequency. A multiobject adaptive spatial sampling approach is presented to construct the fitted model of the electric field intensity generated by a horizontal electric dipole. The [Z] matrix spatial interpolation technology with MoM is described to reduce the computational time needed for the transient characteristics analysis of a horizontal grounding system. The accuracy and computational time of the [Z] matrix interpolation method are compared with those of the direct MoM. The numerical examples show that the [Z] matrix spatial interpolation method can reduce the computation requirement effectively.     

Impedance boundary conditions in a hybrid FEM/MOM formulation

When numerically modeling structures with imperfect conductors or conductors coated with a dielectric material, impedance boundary conditions (IBCs) can substantially reduce the amount of computation required. This paper incorporates the IBC in the finite-element method (FEM) part of a FEM/method of moments (FEM/MoM) modeling code. Properties of the new formulation are investigated and the formulation is used to model three practical electromagnetic problems. Results are compared to either measured data or other numerical results. The effect of the IBC on the condition number of hybrid FEM/MoM matrices is also discussed.     

Analytical evaluation of the MoM matrix elements

Derivation of the closed-form Green's functions has eliminated the computationally expensive evaluation of the Sommerfeld integrals to obtain the Green's functions in the spatial domain. Therefore, using the closed-form Green's functions in conjunction with the method of moments (MoM) has improved the computational efficiency of the technique significantly. Further improvement can be achieved on the calculation of the matrix elements involved in the MoM, usually double integrals for planar geometries, by eliminating the numerical integration. The contribution of this paper is to present the analytical evaluation of the matrix elements when the closed-form Green's functions are used, and to demonstrate the amount of improvement in computation time     

Some convergence considerations in space-domain moment-methodanalysis of a class of wide-band microstrip antennas

The method of moments (MoM) analysis of probe-fed rectangular microstrip patches requires the inclusion of a probe-to-patch attachment mode-expansion function when the substrate thickness d⩾0.02λ, where λ is the free-space wavelength. The results for the input impedance showed increased divergence with measurements when the attachment mode was omitted from the full-wave analysis. The attachment mode can be expressed as an infinite eigenfunction series that increases the fill time of the impedance matrix in an MoM analysis. In an earlier investigation, the infinite eigenfunction series was reduced to a residue series that required one or two terms compared to about 55 terms for the eigenfunction series. In this paper, the convergence properties of the eigenfunction and residue series are investigated in view of rigorous MoM analysis. The relative errors resulting from replacing the eigenfunction by the residue series for the attachment mode, are compared by numerically evaluating a class of two-dimensional (2-D) spatial integrals shown to be closely related to the elements of an MoM impedance matrix. Additionally, the computation times for the evaluation of these integrals for the two forms of the attachment mode-expansion function are also included. Based on the superior convergence properties of the residue series for the attachment mode-expansion function, it is mathematically justified that this form can readily be used for analytic reduction of the spatial, reaction integrals from four to 2-D forms. This feature allows further reduction of the fill time of the MoM impedance matrix, suggesting the possibility of developing an efficient space-domain MoM technique for modeling of wide-band microstrip antennas     

任意形状导体目标的极点求解

从理论上求解极点,频域矩量法结合围线积分是通用且准确的方法,但由于计算量大而一直用于处理旋转对称目标的求解.本文提出Chebyshev多项式展开格林函数的方法,把频率因子从电场积分方程中分离出来,实现了多频点矩量法的快速计算,解决了任意形状导体目标极点的求解问题.最后对围线积分方法进行适当改进,并求解了三种目标的极点.     

深腾1800系统MPI并行矩量法分析线天线

针对目前用矩量法分析电大尺寸物体的辐射和散射时,计算量过大,耗时太长,使分析与优化遇到实质性的困难。研究MPI结合矩量法的并行技术,用并行雅可比迭代法求解矩量法矩阵方程的并行实现过程,测试了其在深腾1800系统中并行矩量法的性能。     

Numerical simulation of scattering from rough surfaces: awavelet-based approach

In this paper, a preliminary study is carried out to demonstrate the application of wavelets for improving the computation time and reducing computational memory required for evaluating the statistics of the scattered field from rough surfaces using the method of moments (MoM) in conjunction with a Monte Carlo simulation. In specific, Haar and the first order B-spline wavelet basis functions are applied to the MoM formulation of one-dimensional rough surfaces in order to compare the computation time and sparsity for wavelets in the same family but of higher order. Since the scattering coefficient (the second moment of the backscatter field per unit area) is a gentle function of the surface parameters and the radar attributes, it is demonstrated that a relatively high thresholding level can be applied to the impedance matrix, which leads to a sparser impedance matrix and faster computation time. It is also shown that applying a high threshold level the coefficients of the high-order wavelets would increase out of proportion, however, the effect of these current components averages out when computing the scattering coefficients. The resulting sparse impedance matrices are solved efficiently using fast search routines such as the conjugate gradient method. A systematic study is carried out to investigate the effect of different threshold levels on the accuracy versus computing speed criterion. The computed scattering coefficients are compared to previous results computed using a conventional pulse basis function as well as the existing theoretical solutions for rough surfaces. It is shown that wavelet basis functions provide substantial reductions in both memory requirements and computation time     

Analysis of the electromagnetic inductive response of a void in aconducting-soil background

A lossless dielectric object situated in a lossy dielectric medium (soil) constitutes a void in a conducting background, which can be detected via an electromagnetic-induction (EMI) sensor operating at appropriate frequencies. The electromagnetic character of this void is dependent on the target and soil properties, as well as on the frequency of operation. We utilize the rigorous method of moments (MoM) and the approximate extended-Born technique to model this three-dimensional (3-D) problem. The modeling algorithms are discussed in detail, with a focus on efficient computation of the dyadic Green's function at the frequencies of interest. The MoM results are used to calibrate the accuracy of the approximate extended-Born solution, over a wide range of operating conditions. Furthermore, the computer simulations are used to perform a detailed phenomenological study