基于时频域的卷积神经网络运动想象脑电信号识别方法

针对目前运动想象脑电（EEG）信号识别率较低的问题，考虑到脑电信号蕴含着丰富的时频信息，提出一种基于时频域的卷积神经网络（CNN）运动想象脑电信号识别方法。首先，利用短时傅里叶变换（STFT）对脑电信号的相关频带进行预处理，并将多个电极的时频图组合构造出一种二维时频图；然后，针对二维时频图的时频特性，通过一维卷积的方法设计了一种新颖的CNN结构；最后，通过支持向量机（SVM）对CNN提取的特征进行分类。基于BCI数据集的实验结果表明，所提方法的平均识别率为86.5%，优于其他传统运动想象脑电信号识别方法；同时将该方法应用在智能轮椅上，验证了其有效性。     

频谱分解技术在塔里木盆地北部TH地区碳酸盐岩缝洞型储层预测中的应用

本文从频谱分解技术的基本原理及计算方法入手,根据正演模型,通过对比缝洞型储层与地震反射特征的关系,总结出储层发育的反射特征,相应地精细调整频谱分解处理参数,进而进行预测。通过与钻井资料对比可知,频谱分解技术可以刻画不同频率强反射、弱反射的边界。     

An improved analysis of experimental data from 125 I hot-line autoradiographs: allowing for the effects of background grains

In this paper we suggest a new statistical method of correcting the results of hot-line experiments for the effects of background sources and we use the new method to reassess the adequacy of three probability distributions proposed in the literature for image spread from line sources. The data are from sources labelled with ¹²⁵I in semi-thin resin sections 0·4-0·8 μm in thickness. The new method reveals that two of the models describe the empirical distributions more closely than earlier analysis had suggested, and it confirms an increasing relationship between half distance of image spread and the thickness of the source. However, it also confirms that considerable ‘inter hot-line’ experimental variation remains, even after background correction. This suggests that multiple experiments are needed to produce reliable estimates of half distance.     

对称非线性系统的无交互控制

本文讨论了具有对称性的非线性系统的无交互控制问题。通过定义对称系统的导出分布族的新概念,我们得出了问题可解的条件。我们取得的结果较不具对称性的情况要简单一些。从而说明,对称性不仅使得系统具有较好的结构,而且使得它的其它一些问题的研究得到简化。     

VaR and ES for linear portfolios with mixture of elliptic distributions risk factors

In this paper, we generalize the Linear VaR method from portfolios with normally distributed risk factors to portfolios with mixture of elliptically distributed ones. We treat both the Expected Shortfall and the Value-at-Risk of such portfolios. Special attention is given to the particular case of a mixture of multivariate t-distributions. This is a part of J. SADEFO-KAMDEM PhD Thesis[12] of the Université de Reims, France . It has been presented at the workshop on modelling and computation in Financial Engineering at Bad Herrenalb, Germany May 6-8, 2003. The author is an associate professor at the Department of mathematics, université d’Evry Val d’Essonne.     

Generalization of the Gram-Charlier/Edgeworth Series and Application to Time-Frequency Analysis

We derive a general equation relating probability densities and as special cases we the obtain Gram-Charlier and Edgeworth series. This allows us to generalize these methods and clarify a number of issues pertaining to both probability theory and time-frequency analysis. In particular we show how the Gram-Charlier and Edgeworth series are related to the kernel method of time-frequency analysis. The approach allows us to construct densities that satisfy given constraints such as joint moments or conditional moments. Also, we show that the kernel has to be signal dependent and that to obtain a proper distribution it should be the ratio of two characteristic functions.     

基于地震道时频分析的地层结构解析原理和方法

 地震时频分析是通过对地震信号采用多种数学变换，实现从时频域角度对信号的局部特征进行分解，剖析信号细微特征所对应的地层结构。在应用地震时频分析技术过程中，人们对于这项技术能否真正有效地解决实际地质问题存在许多争论。本文从时频原理综合分析的角度出发，对地震道时频谱中各种频率特性参数和时频旋回直接转化成对地层结构逐级、分步的解析方法进行了较深入的探讨。研究认为时频分析原理不仅是对地震数据进行特殊数学变换，实现对信号的分解，而且充分利用了地层厚度的调谐效应及岩性调谐效应；时频谱与地质结构特征之间的良好响应关系为地震地质精细解释提供了直观的依据。     

On the recent papers by Nagode et al.

The recent work Professor Nagode and colleagues is discussed with respect to the use of mixture Weibull models.     

具有迭代结构的时频分布计算方法

本文讨论了时频分布中的迭代算法的问题.通过选择特殊的计算窗口,时频分布的迭代运算形式能够有效地利用前面数据段的分析结果,从而避免了重复性的运算,使得计算效率得到提高.本文对原有的利用单边窗口的时频分布迭代算法的性能进行了分析,提出了采用对称型窗口的迭代计算形式.与单边窗口相比,双边形式的计算窗口不但可以有效地提高时频表示精度,同时还能够更为准确的表示信号的瞬时频率.文章对各项理论分析结果提供了相应的仿真实验结果.     

分数阶傅立叶变换的多种定义形式及其光学实现系统分析

分数阶傅立叶变换是经典傅立叶变换的广义形式,它同时从时间域和频率域(或空间域)揭示信号特征。本文系统地分析了分数阶傅立叶变换三种定义形式及其所对应的光学实现系统的组成和原理,说明了光学信息处理系统实现分数阶傅立叶变换的有效性.