文档介绍：重庆大学
硕士学位论文
独立分量分析在盲信号分离中的应用研究
姓名:黄灿辉
申请学位级别:硕士
专业:电路与系统
指导教师:刘晓明
20060401
重庆大学硕士学位论文中文摘要
摘要
近年来,信号处理的理论与方法获得了迅速发展。事实证明,信号处理是推
动众学科发展的一个重要基石。独立分量分析技术( ponent
Analysis, 简称 ICA)是信号处理领域发展较晚的一种理论与方法,已迅速成为该领
域内重要的组成部分,且其发展逐渐趋向成熟化与系统化。
本论文对独立分量分析对于盲信号分离的应用进行了研究,主要工作如下:
(1) 搜集、整理、总结了国内外独立分量分析方面的成果和进展,介绍了独立
分量分析的基本理论,并着重讨论了几种常用的独立分量分析算法及其特点。
(2) 介绍了盲信号分离的基本理论,并在最小互信息盲分离算法的基础上提出
了盲信号分离的基于模糊神经网络的独立分量分析方法。
(3) 论文对基于模糊神经网络的 ICA 对于盲信号分离的应用进行了探讨,制定
了可行的统计独立准则、边缘熵和激活函数,详细介绍了模糊神经网络的设计过
程,并对算法的稳定性和收敛性进行了分析,通过计算机实验证实了该 ICA 算法
对于盲信号分离的良好的性能,指标显示代表信号特征的信号的波形和频率基本
与源信号保持一致达到了较好的分离效果。与同类算法的性能比较表明该算法分
离性能较好。
关键词:独立分量分析,盲信号分离,模糊神经网络,多道信号,熵,互信息
I
重庆大学硕士学位论文英文摘要
ABSTRACT
Recently, theories and methods of signal processing are obtained to develop
quickly. In fact, signal processing is an important foundation stone for all kinds of
subjects' developing. ponent analysis (ICA) technology developing is a
later theory or method in fields of signal processing, but it was important that it has
rapidly e a part of constitution of signal processing fields,and its developing
tends gradually to maturity and systematization.
In this paper, the theory and application of ponent analysis has
been studied. The paper consists of following parts:
1. First, the paper reviews systematically the present research situation of
ponent analysis in the world. The basic principles and concepts of
ponent analysis and some algorithms are introduced.
2. Second, a minimum mutual information algorithm based recurrent neural
networks are given after introduced the basic theory about Blind Source Separation. The
main principle, derivation has been presented.
3. Application research of blind signal separation by using ponent
analysis has been studied in this paper. The feasible independence rule, brim entropy
and activation function have been established. The design of fuzzy works has
been introduced particularly. And the performance of the algorithm has been ill