基于Theil不等系数的IOWA算子最优变形组合预测模型

针对传统组合预测模型中各单项预测模型赋权问题的不足,提出一种基于Theil不等系数的IOWA算子最优变形组合预测模型。利用IOWA算子,依据各单项预测模型预测值各时间点精度的不同,赋予不同的权系数,建立最优组合预测模型,并将其应用于变形数据分析与预测。实验结果表明,本文方法的预测结果比各单项模型与传统组合模型的结果更准确。     

一类非线性奇异边值问题解的存在性

证明一类非线性奇异边值问题解的存在性定理,改进了Ｄｕｎｎｉｎｇｅｒ和Ｋｕｒｔｚ的结果,构造具体的方程说明存在性定理中的条件不能去掉,这表明［４］中一结果不真。     

基于数学形态学细化算子的改进Canny算法研究

边缘检测是图像处理领域研究的一个重要内容。本文基于数学形态学改进现有Canny算子。该方法首先用Canny算子进行滤波,然后采用非极大抑制技术,将强边缘图像和弱边缘图像的边缘进行连接,再对提取的边缘利用形态学细化算子细化。针对不同图像采用不同门限比例,可取得不同的边缘提取效果。本文采用峰值信噪比、均方误差、平均绝对误差三个评价指标对边缘检测算法的优越性进行度量。实验分析表明,本文算法优于Sobel算子、Roberts算子、Log算子及传统Canny算法。     

基于Gabor框架合成小波束叠前深度偏移

利用小波束域波场分解和传播在空间及方向上的双重局域性,通过Fourier传播算子进行深度外推波场,完成小波束的传播,最终应用成像条件求取成像值.合成小波束和小束源记录,进行部分小束源叠前偏移成像的数值试验,获得的成像结果表明,小波束叠前深度成像是切实有效的.     

Geometry of risk analysis (morphogenetic system)

We know that risk analysis is based on the estimation of catastrophic events. Now to obtain a good estimation we must create suitable model of the risk. With samples of data we can use different mathematical models as linear or non linear regression methods, neural network methods, genetic methods and more recent hybrid models consisting of the information diffusion approximate reasoning method and back propagation neural network. In this paper, we present a new frame-work denoted morphogenetic system to study risk analysis. We show that morphogenetic system is a generalization of the tensor calculus used to formalize the non Euclidean geometrical representation. Now, we take suggestion from the electrical circuit analysis to create the morphogenetic model for the estimation. In the electrical circuit, we have the vector of voltages and vector of currents. The relation between voltages and currents is given by Ohms rule and the impedance matrix. The relation is a multi input and multi output (MIMO) system given a set of samples for voltages and currents we can estimate the impedance matrix of the circuit. Now, we know that the electrical circuit is controlled by a limited set of voltage or current sources. The aim of this papers is to give the rule by which from arbitrary voltages in the circuit we compute the voltage sources and also the associated current sources. We also give the method to compute the currents by the current sources and also the voltages by the impedance matrix. All the process from arbitrary voltages, voltage sources, current sources, current in the circuit and voltage in the circuit is denoted the projection operator in the morphogenetic system. Electrical circuits are a prototype of the morphogenetic system. We show that fuzzy inferential processes can be modeled by morphogenetic systems to improve models to estimate risks.     

The spheroidal fixed–free two-boundary-value problem for geoid determination (the spheroidal Bruns' transform)

The target of the spheroidal Gauss–Listing geoid determination is presented as a solution of the spheroidal fixed–free two-boundary value problem based on a spheroidal Bruns' transformation (“spheroidal Bruns' formula”). The nonlinear spheroidal Bruns' transform (nonlinear spheroidal Bruns' formula), the spheroidal fixed part and the spheroidal free part of the two-boundary value problem are derived. Four different spheroidal gravity models are treated, in particular to determine whether they pass the test to fit to the postulate of a level ellipsoidal gravity field, namely of Somigliana–Pizzetti type. Received: 4 May 1999 / Accepted: 21 May 1999     

错格傅里叶伪谱微分算子在波场模拟中的应用

本文在传统傅里叶微分矩阵的基础上,对原始微分算子进行改进,引入了错格微分算子的傅里叶伪谱方法．尽管该方法增加了一些计算量,但却极大地提高了计算精度和稳定性．而且,该方法将微分计算过程由传统的傅里叶变换转换为一般的矩阵矢量乘积,大大降低了微分求解过程的复杂程度．在均匀介质中,将错格伪谱微分算子计算的结果和解析解进行比较,结果表明本文算子几乎达到了解析解的精度．而在分层均匀介质中的实验结果同时显示,该方法精度高、稳定性好,是一种研究层状介质中地震波传播的有效数值方法．     

基于立体图像的牙体重定位分析

介绍了一套对真人牙体进行立体定位、跟踪的方法和系统。在该系统中,运用了计算机视觉的方法解算牙体运动过程中的旋转、偏移量,从而为下一步的医学研究提供依据。实验结果表明,该方法具有直观、简便、快速、精度高等特点,还可推广到其他应用领域。     

优化15点频率-空间域有限差分正演模拟

频率域正演是频率域波形反演的基础,有效快速的正演差分格式可以保证反演结果的精度和效率.本文以用较小的系数矩阵带宽来高效地压制频域正演频散为目标,综合利用加权平均算子、平均加速度项和优化系数三种方法,提出了优化15点差分格式;并且采用压缩存储方式来存放大型系数矩阵,极大地缩小了内存使用量;进而结合最佳匹配层(PML)边界条件,明显地压制了边界反射;最后,通过与前人方法的对比验证,证实了本方法可以在不明显增加计算量的情况下,较好地压制频散.     

基于扩展集合操作的频繁项集挖掘算法研究

对传统集合操作进行了扩展,提出了基于扩展集合操作的最大频繁项集生成算法FIS-ES,并从理论上对算法的复杂度进行了详细的分析。实验表明,在最小支持度较小的情况下,FIS-ES算法比Apriori算法具有更快的挖掘速度、更少的空间占用等优点,与Apriori有很好的互补性。