Second-order random wave solutions for interfacial internal waves in N-layer density-stratified fluid

This paper studies the random internal wave equations describing the density interface displacements and the velocity potentials of N-layer stratified fluid contained between two rigid walls at the top and bottom. The density interface displacements and the velocity potentials were solved to the second-order by an expansion approach used by Longuet-Higgins （1963） and Dean （1979） in the study of random surface waves and by Song （2004） in the study of second- order random wave solutions for internal waves in a two-layer fluid. The obtained results indicate that the first-order solutions are a linear superposition of many wave components with different amplitudes, wave numbers and frequencies, and that the amplitudes of first-order wave components with the same wave numbers and frequencies between the adjacent density interfaces are modulated by each other. They also show that the second-order solutions consist of two parts： the first one is the first-order solutions, and the second one is the solutions of the second-order asymptotic equations, which describe the second-order nonlinear modification and the second-order wave-wave interactions not only among the wave components on same density interfaces but also among the wave components between the adjacent density interfaces. Both the first-order and second-order solutions depend on the density and depth of each layer. It is also deduced that the results of the present work include those derived by Song （2004） for second-order random wave solutions for internal waves in a two-layer fluid as a particular case.     

随机权函数非线性回归模型的异方差统计分析

在回归分析中，随机误差是否存在方差齐性是理论与实际工作者都十分关心的问题，方差齐性假设并不总是正确的，在线性和非线性回归中关于异方差的诊断问题已有许多讨论（[1],[2],[4],[5]）。本文在韦博成（1995）讨论了加权非线性回归模型的基础上，用随机系数的方法，讨论随机权函数非线性回归模型中的异方差检验问题，得到了方差齐性检验的似然比统计量和score统计量，同时，当模型存在异方差时，本文给出了估计方差的一种方法。     

事后分层方法及其在交通运输抽样调查中的应用

本文介绍事后分层技术，说明了它不仅可用于简单随机抽样，而且可以用于放回ＰＰＳ抽样。给出了事后分层估计量的形式，证明了它的无编性，并给出了它在交通运输抽样调查中的应用。     

K—半分层空间是闭映射的不变量

本文证明了：若Ｘ是Ｋ－半分层空间，ｆ是闭映射，则ｆ（Ｘ）是Ｋ－半分层空间，这一结果推广改进了Ｌｕｔｚｅｒ与高国士的有关结果。     

APON系统ONT中AAL5CPCS子层的实现

本文介绍了APON的系统结构和网络分层，并根据AAL5 CPCS子层实现的功能及DSP的优良特性，给出了APON系统ONT中AAL5公共会聚子层的一种实现方案，即应用DSP实现CPCS子层的各项功能。     

连续调查的整群抽样

本文讨论了连续调查中应用普查方法时的整群抽样和分层整群抽样。     

非均匀散射层矢量辐射传输(VRT)方程高阶散射解的迭代法

将散射介质层在z轴方向划分成薄层，用薄层的一阶散射强度、Fourier变换和迭代方法求解散射介质整层的矢量辐射传输(VRT)方程的高阶散射解.该方法将一阶散射与高阶散射迭代结合起来，计算公式简明，可计算高阶迭代解，计算时间少.计算结果与一层均匀散射介质的VRT方程一阶Mueller矩阵解、半空间均匀散射介质二阶Mueller矩阵解、以及离散坐标-特征值特征矢量法的VRT热辐射的数值解作了全面的比较.提出并讨论了非均匀散射层主动与被动VRT方程的高阶解.本计算程序可以通用于非球形粒子多层结构及非均匀介质的散射和热辐射计算. 关键词： VRT方程 分层 迭代解     

Laboratory Studies of the Stratified Rotating Flow Passing over an Isolated Obstacle

We study the flow of a density-stratified fluid passing over an isolated obstacle,using towing-tank experiments.Our special concern is the response of the flow with different Froude numbers passing over a three-dimensional obstacle.A series of experiments of the stratified rotating flow passing over an isolated obstacle was carried out with the towering -tank controlled by the similarity laws and dynamic non-dimension parameters.These experiments show that the Froude number is a very important parameter,and the lee wave and the eddy structure experiments show that the Froude number is a very important parameter,and the lee wave and the eddy structure appear simultaneously under an appropriate conditions.The effect of rotation on the lee wave is mainly to change wave amplitude,particularly to restrain the development of the lee wave and to promote the formation of an eddy.     

偏最小二乘回归的贝叶斯正则化神经网络集成模型在证券分析预测中的应用

神经网络集成技术能有效地提高神经网络的预测精度和泛化能力,已经成为机器学习和神经计算领域的一个研究热点.利用Bagging技术和不同的神经网络算法生成集成个体,并用偏最小二乘回归方法从中提取集成因子,再利用贝叶斯正则化神经网络对其集成,以此建立上证指数预测模型.通过上证指数开、收盘价进行实例分析,计算结果表明该方法预测精度高、稳定性好.