基于稀疏盲源分离的J波提取技术研究

发布时间：2018-01-12 01:18

本文关键词：基于稀疏盲源分离的J波提取技术研究　出处：《太原理工大学》2015年硕士论文　论文类型：学位论文

【摘要】：冠状动脉（冠脉）疾病严重威胁着人们生命健康，能够对高危人群做出及时准确的预测评估极为重要。室性心动过速会有生命危险，并且患者的心电图(ECG)上形成了J波。J点偏移形成的J波，由于其幅值很小，与正常心电信号混叠，想要从临床上的观测信号中看到J波，不是非常容易。如何将可能隐藏的J波从观测信号中提取出来，并进行研究与分类，为临床上冠心病的诊断提供参考依据，成为医学界有价值的热点问题。因此，要解决这个问题，盲源分离（Blind Source Separation，BSS）必是首选。目前，国内外对稀疏性很强的信号做了大量研究，并提出了很多可实现盲源分离的算法，但是，当正常心电信号混合了J波之后，由于源信号不充分稀疏，利用传统的稀疏成分分析（Sparse ComponentAnalysis，SCA）算法不能准确的估计混合矩阵，以便从观测信号中分离出J波。因此，当源信号不充分稀疏时，估计混合矩阵和恢复源信号成为盲源分离技术发展的一次大跨越，这个问题也亟待解决。针对源信号不充分稀疏的情况，本文系统地研究了盲源分离的基本理论，传统SCA方法估计混合矩阵和实现信号盲分离的原理，在传统K-means算法和反馈稀疏成分分析方法能够分离出一个目视效果很好并接近源信号的基础上，提出了反馈部分稀疏成分分析算法（FPSCA），源信号存在频谱重叠，满足部分稀疏性，定位重叠部分的极值点，并搜索其两侧幅值相同的点位，使得混合矩阵能够准确的估计出来。或者先将稀疏点筛选出来，构造比例矩阵来估计混合矩阵，在估计出混合矩阵的一列之后利用FPSCA准确的估计混合矩阵。实验结果证明，在源信号不充分稀疏的情况下，混合通道系数能得到很好的估计，J波能够准确的提取出来。
[Abstract]:Coronary artery disease is a serious threat to people ' s health . It is very important to make timely and accurate prediction and evaluation of high - risk population . J - wave is formed on electrocardiogram ( ECG ) of patients . It is not very easy to find J - wave from observation signal . It is a valuable hot issue for diagnosis of coronary heart disease . Therefore , blind source separation ( BSS ) is the first choice to solve this problem . However , when the source signal is not sufficiently sparse , it is estimated that the mixed matrix and the recovery source signal become a big leap in the development of blind source separation technology when the source signal is not sufficiently sparse . Therefore , the problem also needs to be solved . In this paper , the basic theory of blind source separation is studied systematically in this paper . The basic theory of blind source separation is systematically studied . The traditional SCA method is used to estimate the mixed matrix and realize the blind separation of the signal . In the traditional K - means algorithm and the feedback sparse component analysis method , it is possible to separate out the extreme points of the partial sparsity , locate the overlapping part and estimate the mixed matrix by FPSCA . The experimental results show that the mixed channel coefficient can be estimated well under the condition of insufficient source signal . The J wave can be accurately extracted .

【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：R540.4;TN911.7

【参考文献】