基于VB与MatrixVB的最优分类超球面实现

提出了一种基于共形几何代数最优分类超球面的表示方法;讨论了运用共形几何代数理论来构造最优分类超球可分问题的可行性和简便性;介绍了基于共形几何代数的分类超球面几何表示,并用此表示将二类最优分类超球面的可分问题转化为二次规划的训练学习问题,该算法保留了最大分类间隔理论的优点,将二类最优平面可分推广到最优超球可分。另外针对VisualBasic数值计算能力的不足,不利于系统开发,介绍了基于VB和MatrixVB实现最优分类超球面,该方法将Matlab的强大计算功能与VB的Windows用户界面的开发优势结合起来,充分发挥了各自的特点．缩短了软件的开发周期。软件测试结果表明,计算方法正确,计算速度快,系统资源消耗少,操作简便易行,能满足数据分类的要求。     

基于共形几何代数的圆拟合方法实现

提出一种基于共形几何代数的圆拟合算法,通过将欧式空间嵌入到共形空间,圆可以表示为向量形式,共形空间中点与圆的内积,则表示了点与圆的位置关系.可以证明,通过附加两个变量,圆就可以表示为平面,用圆代替超平面作为决策面,可以降低某些问题的计算复杂度.针对Visual Basic数值计算能力的不足,不利于系统开发;介绍基于共形几何代数圆拟合算法及其基于VB和Matrix VB的实现,该方法将Matlab的强大计算功能与VB的Windows用户界面开发方面的优势结合起来,充分发挥了各自优势,缩短了软件的开发周期.软件测试结果表明,计算方法正确,计算速度快,系统资源消耗少,操作简便易行,能满足数据拟舍的要求.     

模糊C-球壳聚类算法的研究

通过对基于同一目标函数的两种模糊C-球壳(FCSS)聚类算法性能的比较与分析,提出一种新算法。该算法收敛速度快,聚类结果准确。     

基于半监督信息的截集式可能性C-均值聚类算法

截集式可能性C-均值(C-PCM)聚类算法将截集概念引入可能性C-均值(PCM)聚类算法中,明显改善了PCM的聚类中心重合问题,并能够对噪声和奇异点的数据进行有效聚类,但该聚类算法对小目标数据聚类时仍然存在聚类中心偏移的问题。针对此问题,该文将半监督学习机制引入C-PCM的目标函数中,通过部分先验信息来指导聚类过程,提出半监督截集式可能性C-均值(SS-C-PCM)聚类算法。为了提高彩色图像的分割效率和分割准确率,将差分进化超像素(DES)算法获得的图像空间邻域信息融入SS-C-PCM目标函数中,并利用彩色直方图重构目标函数,以降低算法的计算复杂度,进而提出基于差分进化超像素的半监督截集式可能性C-均值(desSS-C-PCM)聚类算法。通过人造数据和彩色图像分割的仿真并与多种相关算法进行对比,表明该文算法能够有效改善小目标数据的聚类效果,提高算法的执行效率。     

基于半监督信息的截集式可能性C-均值聚类算法

截集式可能性C-均值(C-PCM)聚类算法将截集概念引入可能性C-均值(PCM)聚类算法中,明显改善了PCM的聚类中心重合问题,并能够对噪声和奇异点的数据进行有效聚类,但该聚类算法对小目标数据聚类时仍然存在聚类中心偏移的问题.针对此问题,该文将半监督学习机制引入C-PCM的目标函数中,通过部分先验信息来指导聚类过程,提出半监督截集式可能性C-均值(SS-C-PCM)聚类算法.为了提高彩色图像的分割效率和分割准确率,将差分进化超像素(DES)算法获得的图像空间邻域信息融入SS-C-PCM目标函数中,并利用彩色直方图重构目标函数,以降低算法的计算复杂度,进而提出基于差分进化超像素的半监督截集式可能性C-均值(desSS-C-PCM)聚类算法.通过人造数据和彩色图像分割的仿真并与多种相关算法进行对比,表明该文算法能够有效改善小目标数据的聚类效果,提高算法的执行效率.     

基于改进KFCM聚类的数据关联算法

针对传统的基于模糊C-均值(FCM)聚类的数据关联算法存在的缺陷,提出了一种基于改进核函数模糊C-均值(KFCM)聚类的数据关联算法。该算法以改进的KFCM聚类为基础,通过放宽KFCM聚类的约束条件来增强系统的鲁棒性,并引入信息熵自动确定目标数以作为数据关联的前期准备,再将改进的KFCM聚类算法引入JPDA算法,通过避免对联合事件的概率计算和对确认矩阵的拆分,以实现数据的正确关联和对多目标的实时跟踪。仿真结果表明算法有效可行。     

一种粒子群算法与FCSS相结合的模糊球壳聚类算法

模糊球壳聚类算法（FCSS）广泛应用于模式识别与机器学习领域。由于其采用传统的基于梯度法和交替寻优策略求解模型,对初始值比较敏感,往往只能得到模型的局部极值点,从而影响聚类的效果,甚至使所采用的方法失效。本文将现代全局优化方法之一的粒子群优化算法与模糊球壳聚类算法（FCSS）相结合,利用粒子群算法良好的全局收敛能力来改善传统聚类算法易于陷入局部极值的缺陷,从而得到一种新的球壳聚类算法（PSO-FCSS）,数值实验表明,新方法对球壳形数据有令人满意的聚类效果。     

基于模糊核C-均值聚类分析的HRRP识别

由于雷达目标及其所处环境的复杂性,导致目标之间的关系往往是非线性的,因此,基于核方法的模式识别方法被广泛应用于雷达目标识别中。在对模糊核C-均值聚类算法深入研究的基础上,提出一种基于模糊核C-均值聚类的高分辨距离像识别算法。该算法针对特征提取后一维距离像数据的特点,采用组合核函数以降低由于数据属性数值过大造成的权重过大对识别效果的影响;同时,算法可以在训练过程中通过有效性函数自适应地确定最佳聚类数目。仿真实验结果表明,基于组合核函数的识别算法同基于传统的高斯核的算法都能有效识别雷达目标,但前者具有更高的目标识别率。     

雷达数据关联中动态加权模糊C-均值聚类算法研究

针对雷达本身及测量的运动目标淹没在大量的杂波中而导致所测数据的不准确问题,文中对雷达测量的大量目标数据进行卡尔曼滤波,以减少数据关联时的计算冗余量;并对模糊C-均值(C-Means)聚类算法进行改进,改进后的算法利用实时目标航迹斜率的变化率对传统的模糊C-Means聚类算法进行动态加权,从而使模糊C-Means聚类算法的目标函数最优化,优化后的目标函数确定的聚类中心更加逼近目标的实际值,从而保证数据关联的准确度,并减少了计算时间,提高算法的效率.仿真实验表明,将文中基于目标航迹斜率变化率动态加权的模糊C-Means聚类算法应用于曲线运动目标的数据关联中,与传统的模糊C-Means聚类算法相比,可以提高数据关联准确度和效率.     

基于类内差异和类间重叠的有效性函数

模糊C-均值(FCM)聚类算法的一个主要问题是需要事先确定聚类的数目,为此定义了类内差异度和类间重叠度来分别度量同一个聚类中数据的相似度和不同聚类间的分离程度,进而基于这两个度量提出一个新的有效性函数用于判定最佳聚类数目。实验结果表明,该有效性函数能有效地判定聚类数目,并且有较好的鲁棒性。     

The application of two-dimensional signal transformations to theanalysis and synthesis of structural excitations observed in acousticalscattering

Acoustic scattering from air-filled, elastic shells submerged in water is an important problem in applied science. The excitations of interest yield a set of physically distinct components to the impulse response of a shell. The components form a natural basis for all signals which can be observed in acoustical scattering experiments from the shell via temporal convolution with some chosen input signal. The Fourier transform (FT) of the impulse response of a shell yields its transfer function, which is also called the form function. We study two types of shells in this paper: a spherical shell, and a finite, ribbed, cylindrical shell with endcaps. Utilizing several different two-dimensional (2-D) signal transformations, we can decompose the response of the shells. The resulting 2-D images allow for a striking visual decomposition of the responses into their distinct components. In the case of the spherical shell, a virtually exact theory exists that allows for analytic synthesis of the shell response into its components. However, for the more complex cylindrical shell, the theory for the direct scattering problem is not nearly so mature. Yet, we can still decompose experimentally-obtained shell responses into their distinct components via signal synthesis techniques applied to the 2-D transforms     

模糊C-球壳聚类算法及其在血液细胞图像中的应用

文章从经典的模糊C均值算法开始通过改变其中相似性的度量形式，介绍了一种模糊C球壳聚类（FCSS）算法。在将该算法应用于细胞显微图像半径统计时，采用基于形态学的图像预处理措施，可以获得FCSS算法中有关原型模式的知识，加快收敛速度并避免随机初始化造成的局部极小问题。     

Scattering by dielectric containers

The radar cross section (RCS) has been derived by an approximate method for a thin dielectric plate and for thin spherical and conical dielectric shells. The computed results for a flat plate and a conical shell are compared with experimental patterns. The computed results for a conical shell are presented in the form of constant RCS contours on the aspect angle-frequency plane.     

Inwards Interweaving of Polymeric Layers within Hydrogels: Assembly of Spherical Multi‐Shells with Discrete Porosity Differences

The unique inwards interweaving morphology of polyamines and polyacids within agarose hydrogels that leads to the formation of striated shells with different porosities within the spherical scaffold is reported. Microcompartments with sophisticated structures are commonly used in drug delivery, tissue engineering, and other biomedical applications. However, a method capable of producing well‐defined, multiporous shells within a single compartment is still lacking. By the alternating deposition of polyallylamine (PA) and polystyrenesulfonic acid (PSS) in 1‐butanol, at equal mass ratios, multiple levels of porosity are generated within an agarose microsphere. Each level of porosity is represented by a well‐defined, concentric shell of interweaving PA and PSS layers. The number, thickness, and porosity of the striated shells can be easily controlled by varying the number of PA/PSS bilayers and the polymer concentration, respectively. The feasibility of utilizing this morphology for the assembly of a multi‐shell porous spherical scaffold is validated by trapping different molecular weight dextrans within different regions of porosity. The unique interaction of polyacids and polyamines in hydrogels represents a facile and inexpensive approach to the development of intricate scaffold architectures.     

纵弯转换超声振动球面聚焦系统聚焦特性研究

该文提出了一种由夹心式纵振动换能器与带孔球壳组成的纵弯转换型球面聚焦系统。应用弹性力学的板壳振动理论和亥姆霍兹-基尔霍夫声场理论分析了新型聚焦系统的振动特征和声场聚焦特征,并通过实验进行了验证。研究结果表明,球壳弯曲振动的辐射声场具有显著的聚焦特性,焦点位置、声压强度、焦区形状受球壳曲率半径的影响。在结构一定的情况下,聚焦的特性受谐振频率的影响,高阶谐振频率的聚焦效果比低阶谐振频率好,但焦点声压低于低阶谐振频率。由于弯曲振动与气体介质的辐射阻抗匹配要好于纵向振动,因此这种新型聚焦系统具有更强的实际应用意义。     

Core–Shell Nanoparticles: Characterizing Multifunctional Materials beyond Imaging—Distinguishing and Quantifying Perfect and Broken Shells

Core–shell nanoparticles (NPs) are amongst the most promising candidates in the development of new functional materials. Their fabrication and characterization are challenging, in particular when thin and intact shells are needed. To date no technique has been available that differentiates between intact and broken or cracked shells. Here a method is presented to distinguish and quantify these types of shells in a single cyclic voltammetry experiment by using the different electrochemical reactivities of the core and the shell material. A simple comparison of the charge measured during the stripping of the core material before and after the removal of the shell makes it possible to determine the quality of the shells and to estimate their thickness. As a proof‐of‐concept two multifunctional examples of core–shell NPs, Fe₃O₄@Au and Au@SnO₂, are used. This general and original method can be applied whenever core and shell materials show different redox properties. Because billions of NPs are probed simultaneously and at a low cost, this method is a convenient new screening tool for the development of new multifunctional core–shell materials and is hence a powerful complementary technique or even an alternative to the state‐of‐the‐art characterization of core–shell NPs by TEM.     

Conformal antenna arrays with E3M3 radiators

Design relationships for computer simulation of radiation patterns of conformal antenna arrays are derived. Results of the comparative analysis of radiation patterns of arrays with spherical, hemispherical, and disk-shaped apertures in which the radiators consist of three electric dipoles and three magnetic dipoles (E3M3) are presented. It is shown that application of E3M3 radiators with proper orientation of their axes in conformal antenna arrays can increase the directive gain in the main lobe.     

Conversion of Fly Ash Cenosphere to Hollow Microspheres with Zeolite/Mullite Composite Shells

A seed‐induced in‐situ hydrothermal conversion technique is proposed to prepare novel hollow microspheres with zeolite/mullite composite shells from fly ash cenosphere (FAC), a solid waste with a hollow structure from coal power stations. Two groups of hollow microspheres were prepared, one with zeolite FAU/mullite composite shells and the other with zeolite LTA/mullite composite shells. The FAC in this study plays dual roles as both the template cores and the aluminosilicate nutrition contributor. The final products inherit the hollow spherical morphology of FAC and possessed bilayered shells, the outer dense shell of zeolite crystals and the inner porous shell of mullite. Such hollow zeolitic materials are expected to have many advantages in applications such as catalysis, adsorption, separation, and as releasing capsules.