基于改进型双正切变换的高阶近奇异性精确积分方法研究

电磁场表面积分方程方法（SIE）中的高阶近奇异性积分是SIE精确求解的关键技术之一,但现有方法主要是处理平面单元建模中的低阶近奇异性问题,目前还没有一种可用于高阶曲面建模中3阶近奇异性的精确稳定积分方法.本文在前期提出的双正切变换方法（DAT）的基础上,针对高阶曲面建模中含有RR/R⁵、R/R⁴和1/R³等形式积分核的近奇异性问题,通过引入指数变换解决了DAT算法在近奇异点与源单元非常接近时算法不稳定的问题,并通过引入形函数变换解决了DAT近奇异点与源单元边界靠近时积分不稳定的问题,形成改进型双正切变换方法（IDAT）.相对于DAT,所提IDAT更稳定高效.所提IDAT不仅可用于曲面单元中的高阶近奇异性问题的精确积分,同时也适用于低阶近奇异积分问题.理论分析与数值算例验证了本文所提方法的精确性与稳定性.     

基于RWG基函数的伽略金法中奇异性积分 的精确快速计算

利用电磁场积分方程的伽略金法求解理想导体电磁散射问题时需要计算奇异性的二重面积分(即4维积分).伽略金法的基函数和检验函数广泛采用RWG(Rao-Wilton-Glisson)矢量基函数.传统上采用奇异值提取技术和Duffy坐标变换法处理该奇异性积分,本文提出了一种更为精确和高效的计算方法,该新方法通过参数坐标变换、相对坐标变换、积分区域分解和广义Duffy坐标变换相结合的技术消除了被积函数的奇异性并降低了原4维奇异性积分的数值积分维数.通过计算实例证明该方法的精确性和高收敛特性.     

三维矢量散射分析中奇异积分的准确计算方法

在用积分方程和矩量法(MM)或快速多极子法(FMM)分析三维矢量散射时,都要对有奇异性的被积函数进行积分。如果直接使用高斯积分,则准确性很低。为了得到准确的积分结果,本文在分析了高斯积分原理的基础上提出了积分区域分割法。此方法将积分区域划分为一个包含奇异点的部分和若干个无奇异点的部分。对无奇异点的部分可直接用高斯积分求解,而对包含奇异点的部分,则可通过简化被积函数,变量代换和加减同阶奇异项等方法获得它的解析表达式。本文将这种方法用到电场积分方程(EFIE)的矩量法中,以角反射器和导电球目标散射特性(RCS)为例,其计算的结果与文献非常吻合。     

三维矢量散射积分方程中奇异性的分析

本文研究了电场积分方程（ＥＦＩＥ）中被积函数奇异性的处理方法,特别是三维矢量散射分析中出现的高阶奇异性,给出了两种解决积分方程奇异性的数值方法。一种方法是计算Ｏ（１／Ｒ）阶奇异积分转移法,另一种方法是为解决Ｏ（１／Ｒ＾２）高阶奇异积分的数值计算问题的,它是通过排除一包含奇点的有限小块,而这一小块区域对积分的贡献为零,从而使积分方程在整个积分域变得数值可积。     

精确稳定求解时域电场积分方程的一种新方法

时域电场、磁场和混合场积分方程已被广泛用来分析散射体的时域散射响应.基于适当的空间积分方法和隐式的时间步进算(MOT)法在求解时域磁场和混合场积分方程时总是稳定的,然而在求解TDEFIE时则是不稳定的.在本文中,时域电场积分方程的非奇异性积分采用标准的高斯求积法来计算;而利用参数坐标变换和极坐标变换将其奇异性积分转换成为可以分区域精确快速计算的非奇异性积分.通过数值实验表明,利用该方法可以非常精确稳定地求解时域电场积分方程,即使是在时间迭代后期也不必采用任何求平均的过程;另外,该方法可以用于任意时间基函数并可以推广到高阶空间基函数的情形.     

半空间格林函数的角谱平面高效计算

将半空间并矢格林函数的索末菲积分用角谱平面积分表示,将原始的索末菲积分路径变换至最陡下降路径,针对场源位于不同区域时的分支割线和支点分布做了讨论,给出了最陡下降路径方程和相应的积分公式.对于可能涉及的支点奇异性,通过增加一条过支点的等相位路径来计算.角谱平面的计算方法有效地简化了SDP方程的推导与积分计算过程;数值计算表明基于最陡下降路径的积分方法可以极大提高半空间格林函数的计算效率.     

精确快速计算时域积分方程中奇异性积分的新方法

该文首先利用参数坐标和广义Duffy坐标变换将时域电场积分方程(TDEFIE)的奇异性积分转换成非奇异性积分,然后根据时间基函数的特点将该积分转换成可以快速精确计算的分区域积分。数值计算实例表明,该方法可以大幅度提高求解TDEFIE的后时稳定性和解的精度,而不必采用任何求平均的过程。该方法适用于任意类型的时间基函数并可方便地推广到高阶曲面拟合和高阶空间基函数情形。     

求解时域电场积分方程的稳定时间步进算法

利用面积坐标变换、相对坐标变换、积分区域分解和广义Duffy坐标变换将时域电场积分方程中奇异性积分(共面、共边和共单结点的场源三角形单元上)转化成可精确计算的非奇异性积分.在不同时间基函数(导数连续和导数不连续)、不同时间步长情况下对比分析了该方法和现有的常用方法计算奇异性积分的精度.计算实例表明:时域阻抗矩阵的精确计算有效地改善了时间步进算法的后时稳定性.     

非均匀介质结构中线天线辐射问题的有限元-边界积分方法

采用有限元边界积分方法,通过把所求解区域分成内外两部分,内场用有限元计算,外场用边界积分计算,分析了线天线辐射问题.边界积分采用伽略金方法处理,并用变换解决了积分方程的奇异性问题,最后给出了仿真结果.     

平面电路分析中奇异函数积分的计算

在利用边界元法分析平面电路问题时,常会遇到一些被积函数在积分区域内具有奇异性的积分。本文提出了求解这类积分的新思路,即先采用适当的变量代换达到分离变量或改变奇异点在积分区域内位置的目的,然后采用奇异点去除法计算出简化后积分的值。该方法不仅能求解数值方法所无法求解的积分,而且由于尽可能地采用了解析方法,因而具有计算速度快,精度高等特点,该方法为奇异积分的计算提供了一种行之有效的手段。     

RWG 基伽略金矩量法自/ 互阻抗计算中两个关键问题的评述

为了快速获得RWG基伽略金矩量法自/互阻抗精确值,一般采用数值方法与解析技术相结合的求积策略。求积策略中的数值积分方法采用三角形高斯求积,而解析技术则普遍采用奇异值提取技术。针对这两个关键问题在应用过程中容易忽视的几个细节进行了评述,包括三角形高斯求积规则选取、求积公式应用条件以及奇异性积分被积函数改造等。采用新近提出的奇异性积分精确快速算法对自/互阻抗计算涉及的两类积分进行了推导计算,同时给出了场、源三角形完全重合和具有公共边两种情形下,采用常规奇异值提取技术和精确快速算法对这两类积分的计算结果。     

Evaluation and integration of the thin wire kernel

New approaches for numerically computing the thin wire kernel and wire potential integrals are presented. The singular behavior of the kernel integral is removed by transforming the integration variable to produce a smooth integrand. Subsequent integration of the kernel to obtain potential integrals uses quadrature schemes catering to its behavior. This technique allows standard algorithms for numerical quadrature to be used with updated integration weights that account for the transformed behavior, obviating the need for singularity subtraction techniques. The result is a procedure for evaluating the potential integrals that is independent of the basis functions.     

A Legendre Approximation Method for the Circular Microstrip Disk Problem

The quasi-static solution for the circular microstrip disk is studied using a GaIerkin solution to the Fredholm integral equation of the first kind derived by using the Green's function approach. The basis functions are modified Legendre polynomials combined with a reciprocal square root to provide the correct singularity in charge density at the edge of the disk. The integrals involving the singular part of the Green's function are evaluated exactly, the remainder by using Gaussian quadrature. The method is compared in computational efficiency with recent methods based either on a Galerkin approach in the spectral domain, or the use of dual integral equations. Numerical results are given for charge distribution and capacitance; they are compared to exact results and those obtained by others, and the limitations of those methods are discussed. Closed form expressions are given for the capacitance of a disk based on two simple charge distributions.     

Time-domain fields exterior to a two-dimensional FDTD space

A transformation algorithm for the near-zone and far-zone fields exterior to a two-dimensional (2-D) finite-difference time-domain (FDTD) field lattice has been developed entirely in the time domain. The fields are found from a surface integration of the convolution of the time derivative of equivalent currents and charges along a contour that encloses the scatterer or radiator of interest. The kernel of the convolution integral has a square-root singularity for which an efficient numerical integration rule is presented. Using this technique, a very accurate solution is obtained; however, convolution integrals are computationally expensive with or without singularities. As an alternative, a rapidly convergent approximate series expansion for the convolution integral is presented, which can be used both in the near and far zone. Results using the new 2-D transform are compared with analytical expressions for the fields generated by a modulated Gaussian pulse for TE and TM line sources. In addition, the 2-D transform solution for the near-zone fields scattered from an open-ended cavity for a TE incident modulated Gaussian pulse plane wave is compared against a full-grid FDTD solution for accuracy and efficiency. The 2-D transform far-zone fields are compared against an alternative technique, which uses a double Fourier transform to perform the convolution in the frequency domain     

Accurate and efficient numerical integration of weakly singular integrals in Galerkin EFIE solutions

A Galerkin descretization of the electric field integral equation for perfectly conducting surfaces using Rao-Wilton-Glisson (1982) basis functions requires the numerical evaluation of integrals with singular kernels over triangular regions. These singularities have been traditionally handled by utilizing a "singularity extraction" procedure to produce a regular integral and an analytic function to replace the original singular integral. A new approach is presented here in which the four-dimensional (4-D) weakly singular integrals unique to the Galerkin Rao-Wilton-Glisson electric field integral equation solution for perfectly conducting surfaces are transformed into integrals with regular integrands. The transformations allow some of the integrations to be performed analytically, in some cases reducing the original 4-D integral into a 1-D numerical integration. The accuracy and convergence properties of the new method are demonstrated by evaluating the scalar potential function over a unit triangle.     

导体线面连接问题中奇异函数积分的计算

在线面连接问题中，电流展开函数包含体展开函数、线展开函数和连接点展开函数三类。求解电场积分方程的积分项是电流基函数及其散度分别与自由空间格林函数的乘积，由于连接点展开函数含有一个奇异点，所以被积函数中含有两个奇异点。本文通过积分变换消除了奇异点，并将二重积分化为一重积分，使计算精度得到提高。计算实例验证了本文方法的正确性。     

General singularity extraction technique for reflected Sommerfeld integrals

A general technique for singularity extraction from reflected Sommerfeld integrals (SIs) in frequency domain is presented. The essence of the technique is an analytical evaluation of all the singular and slowly convergent terms in SIs for reflected potentials and fields. This is done starting from two basic integrals. Up to two terms for Sommerfeld potential integrals and up to three terms for Sommerfeld field integrals are extracted. The remaining well-convergent parts of integrals are evaluated by numerical integration along the real axis. A new type of singularity extraction is applied to Sommerfeld's g_Azx term, yielding the generalized Foster–Lien integral. Accuracy and efficiency of the method are illustrated by a numerical example.     

Application of a point-matching MoM reduced scheme to scatteringfrom finite cylinders

One of the most common methods for the solution of three-dimensional (3-D) scattering problems is the electric-field volume integral equation numerically solved by the application of the method of moments (MoM)-usually the point-matching version. Although simple to formulate, it shows inherent difficulty and complexity because of the 3-D integrals appearing in the interaction matrix elements and of the singularity of the dyadic Green's function (DGF) present in the computation of the self-cell elements. In this paper, a transformation method is presented, which in the case of the point-matching MoM, both reduces the 3-D integrals to two-dimensional (2-D) ones, and also eliminates the need of separate treatment of the singularity while maintaining the same degree of approximation. Comparison to published results is made for the case of scattering by a finite dielectric cylinder. Further examples are presented for scattering by layered dielectric cylinders and lossy cylindrical shells excited by uniform plane waves     

采用分部外推BCGS-FFT方法快速求解电磁散射问题

为了快速求解电磁散射问题中具有震荡性、奇异性、慢收敛性的索末菲积分,提出了一种利用分部外推算法加速索末菲尾部积分计算,并结合稳定双共轭快速傅里叶变换(stabilized biconjugate gradient fast Fourier transform,BCGS-FFT)算法求解电磁散射问题场分布情况的新方法. 首先给出电场积分方程(electric field integral equation, EFIE)的表达形式,且在求解过程的索末菲积分中应用一种便捷的椭圆积分路径来最小化索末菲积分的震荡性与奇异性,在索末菲尾部积分使用Levin分部外推法来提高积分收敛速度,以此来快速填充并矢格林函数矩阵. 然后对新方法进行了多种数值实验,验证算法的精确度,并对比了新方法与传统BCGS-FFT方法的计算效率,发现在保持相同计算精度的条件下,新方法可节省20%～37%的计算时间. 该方法能应用于复杂散射体嵌入多层空间的电磁散射计算,为快速求解目标区域的电磁散射场提供了一种新的方法.