任意截面波导的小波矩量法分析

用离散小波矩量法推导出了任意截面波导的特征值方程,并以常见规则波导为例,对其截止波数进行数值计算,所得结果与理论值吻合良好,从而使小波矩量法求解波导问题的可行性和有效性得以验证;文中分别用矩量法和小波矩量法对矩形波导主模的截止波长进行计算,通过对比其运算时间和收敛速度,表明在保证精度的前提下,小波矩量法能有效提高运算速度。     

小波矩量法分析环形微带缝隙天线*

将矩量法和小波变换理论结合起来,应用于求解电磁场积分方程,分析计算了环形微带缝隙天线磁流分布、驻波、阻抗特性和辐射方向图,通过用小波基代替一般的分域基矩量法中的权函数和基函数,使矩量法中与算子相对应的阻抗矩阵变成为大量元素为零的稀疏矩阵,减小了数值方法对存储量的要求,在达到允许精度的前提下,使计算量显著降低。文中给出计算实例,验证该方法,计算结果与用一般矩量法计算结果和实际测试结果相比,吻合较好。     

基于小波矩量法的平面螺旋电感的电磁辐射研究

 平面电感实现了电感器件的小型化和片式化.当工作频率达到射频或微波段时,应考虑平面电感线圈的辐射和散射对其周围电路的影响.本文采用小波矩量法研究了PCB(printed circuit board)平面螺旋电感的电磁辐射特性.采用区间小波作为矩量法中的基函数和检验函数,计算了电感中的电流分布,给出了辐射方向图.与传统矩量法相比,小波矩量法可以使系数矩阵稀疏化,从而节省了计算机资源和计算时间.     

小波矩量法计算导电平面上的三维腔体散射

该文采用小波矩量法求解导电平面上的三维腔体散射,利用小波变换稀疏化矩量矩阵,加速其求逆,减少了腔体散射的总体计算时间。通过具体的算例,表明小波矩量法在计算三维散射问题中是非常有效的。     

用条件预优加速小波矩量法的计算

本文在小波矩量法中利用阻抗矩阵稀疏的特点,用条件预优法快速求解电磁散射问题,讨论了矩量法的采样密度对预优方法的影响。通过二维理想导电柱体散射的例子,表明该方法十分有效地加速了小波矩量法的计算,并且比文献（1）的方法更显著地减少了计算时间。     

图像的小波矩

本文从一般特征矩的定义入手,导出极坐标系下图像的小波矩,进而给出对图像进行多分辨小波变换后,利用任一尺度下的逼近系数构造小波矩和基于小波模极大值边界图像构造小波矩两种方法.最后,经实验验证了本文构造图像小波矩方法的有效性.     

周期小波在电磁散射中的应用

讨论周期小波与矩量法相结合的方法在二维电磁散射计算中的应用，以周期小波作为基函数和权函数，并利用ｃｏｉｆｌｅｔ小波的消失矩性质，求解速度较快，结果非常精确。本文分别用小波矩量法、传统的矩量法及广义多极子方法计算了一个无限长方柱在平面波照射下的电流分布，并与有关资料上的结果进行比较，吻合得很好。     

基于多尺度分析的小波不变矩

本文结合小波多尺度分析和不变矩的优点,通过坐标变换和几何规范化等预处理,提出了一种能提供多尺度分析且具有平移、尺度和旋转不变性的矩特征描述子-小波不变矩,并给出了其不变性证明.讨论了不同类型和阶数的小波函数对小波不变矩识别性能的影响.实验结果表明,由合适消失矩阶数的、对称紧支集双正交小波函数得到的小波不变矩具有最佳的分类特性.     

基于小波变换的多项式相位信号检测

N阶消失矩的小波能够消去n-1阶多项式。本文用具有不同消失矩小波,依次消去多项式相位信号不同阶次的相位向,实现单个多项式相位信号的检测。仿真结果表明,这一算法能够有效地检测多项式相位信号。     

基于小波结构矩的图像识别算法

提出了一种基于小波结构矩的具有平移、旋转、缩放不变性的新型图像识别方法.小波结构矩是在小波矩的基础上通过改变图像函数的结构即几何矩的密度得到的.该算法结合了小波和结构矩的优点,不但实现了对图像特征的精细把握还增大了相似图像之间的距离.采用三次样条函数作为母小波,有效提取了图像的全局特征和局部特征.实验证明,小波结构矩比改进的Hu矩和结构矩具有更高的识别率,目前该算法已经成功运用到全自动金丝球焊机图像识别系统.     

A theorem on the moment methods

The inner product involved in the moment methods is usually an integral, which is evaluated numerically by summing the integrand at certain discrete points. In connection with this inner product, a theorem is proved, which states that the overall number of points involved in the integration must not be smaller than the number of unknowns involved in the moment method. If these two numbers are equal, a point-matching solution is obtained, irrespective of whether one has started with Galerkin's method or the least squares method. If the number of points involved in the integration is larger than the number of the unknowns, a weighted point-matching solution is obtained.     

Incompleteness of the periodic moment method

The periodic moment method predicts zero mutual impedance between expansion and test dipoles that are perpendicular. A new term has been added for this case, with nonzero results that agree precisely with spatial domain mutual impedance summations.<>     

Error analysis of the moment method

Because of the widespread use of the Method of Moments for simulation of radiation and scattering problems, analysis and control of solution error is a significant concern in computational electromagnetics. The physical problem to be solved, its mesh representation, and the numerical method all impact accuracy. Although empirical approaches such as benchmarking are used almost exclusively in practice for code validation and accuracy assessment, a number of significant theoretical results have been obtained in recent years, including proofs of convergence and solution-error estimates. This work reviews fundamental concepts such as types of error measures, properties of the problem and numerical method that affect error, the optimality principle, and basic approximation error estimates. Analyses are given for surface-current and scattering-amplitude errors for several scatterers, including the effects of edge and corner singularities and quadrature error. We also review results on ill-conditioning due to resonance effects and the convergence rates of iterative linear-system solutions.     

On the reconstruction aspects of moment descriptors

The problem of reconstruction of an image from discrete and noisy data by the method of moments is examined. The set of orthogonal moments based on Legendre polynomials is employed. A general class of signal-dependent noise models is taken into account. An asymptotic expansion for the global reconstruction error is established. This reveals mutual relationships between a number of moments, the image smoothness, sampling rate, and noise model characteristics. The problem of an automatic (data-driven) section of an optimal number of moments is studied. This is accomplished with the help of cross-validation techniques     

A macro-element compression for the moment method

A new approach based on QR compression of the off-diagonal blocks of the impedance matrix arising in integral equations solutions of electromagnetic (EM) scattering problems, previously implemented for finite and irregular gratings, is extended here to an arbitrarily shaped obstacle. The main tool consists in dividing the entire surface scatterer into subdomains and in assembling the global impedance matrix as a combination of the macromatrices associated to each interaction of any patch on the other. The correlation between compression rate and accuracy of field computation is drawn up. An experimental study is performed to optimize the method relatively to the two competing parameters: accuracy and cost. Finally, it is shown that this method works equally well in both the high and the low frequency domain and also with an inclusion of a small detail in a large object     

天堂一瞬：马尔代夫蜜月之旅

在印度洋宽广的蓝色海域中,有一串如同被白沙环绕的绿色岛屿——马尔代夫群岛,被认为是地球上最后的乐园。有人形容马尔代夫是上帝抖落的一串珍珠,也有人形容是一片碎玉,这两种形容都很贴切,白色沙滩的海岛就像一粒粒珍珠,而珍珠旁的海水就像是一片片美玉。     

Generalized moment estimators for the Nakagami fading parameter

Moment-based estimation of the Nakagami-m fading parameter is studied. A family of new, generalized, estimators, based on real (integer and possibly noninteger) sample moments, is derived. A new integer-moment estimator that outperforms known integer-moment estimators is presented. The new moment-based estimators are asymptotically unbiased and efficient     

Least-squares moment matching reduction methods

It is shown how the two apparently different approaches to modelling reduction by least-squares time moment matching give identical results. This property also enables a clear understanding of how the method actually approximates the system. Computational savings are seen to be made by adopting the two-stage approach