基于混沌理论的心音信号非线性动力学分析
发布时间：2019-01-03 20:40
【摘要】：心音是人体重要的生理信号之一，能反映心脏及大血管的机械运动状况，是临床评估心脏功能状态的最基本无创而便捷的方法。由于生命是最复杂的非线性动力系统，而心脏是生命循环系统的核心，这就决定了由心脏振动所产生的心音信号的非线性及复杂性。一直以来，人们为了对心脏这个复杂的系统进行简化及抽象，而对其建立一个理想的线性模型，并且用时域、频域、时频转换等方法对该线性系统进行分析和处理。但是，半个世纪以来，人们发现用线性的方法进行分析并不足以研究本质上为非线性的生命活动。由于混沌作为非线性系统的一种极为重要的运动形态，可以很好地揭示非线性过程内在随机性所具有的特殊规律性，从而本课题拟从混沌理论的角度对心音信号进行分析，从本质上更深入地认识心音信号的内在特征规律，，以期从一个全新的角度实现基于心音信号的心脏疾病的计算机辅助诊断。 为了提高心音信号的识别精度和分类准确性，采用小波包分析及混沌理论结合的方法对心音信号进行特征提取及分类识别。与小波变换相比，小波包具有更强的时频分辨力，从而能够提取原始信号局部更精细的时频信息。一方面从时频角度采用小波包对心音信号进行分析，利用小波包将心音信号分解成不同频段，再对分解后的频段作能量特征的提取；另外，将小波包分解的心音信号分量中能表征心音信号特征的信号分解出来，对其进行混沌分析，包括定性及定量分析，其中定性分析包括心音信号相图及递归图，定量分析包括关联维数、最大Lyapunov指数等混沌特征参量；然后将小波包分解的子带能量特征和混沌特征参数结合构成心音信号特征参数矢量，再通过遗传算法分析心音信号小波包各频带能量特征以及混沌特征参数，选取了能够表征心音信号的最优特征矢量；最后采用支持向量机(SVM)作为分类器，以心音信号特征矢量作为输入，从而实现心音信号的自动分类识别。 通过设计的心音信号采集系统，对临床采集的正常及几类异常心音信号，如早搏心律不齐、二尖瓣狭窄、第一心音分裂、主动脉关闭不全及室间隔缺损等异常心音，采用本文所述方法进行测试。结果表明，正常及异常心音信号的混沌定性定量特征都具有显著性差异，其中异常心音信号的关联维数及最大Lyapunov指数都较正常心音信号高，说明异常心音信号具有较高的复杂度。结合小波包能量及混沌特征的心音信号能够获得较高的识别率，说明混沌特征对于心音信号非线性特征的揭示具有重要的作用，为后期心音信号的诊断及心音非线性本质的研究奠定了基础。
[Abstract]:Heart sound is one of the important physiological signals of human body, which can reflect the mechanical movement of heart and large vessels. It is the most basic noninvasive and convenient method to evaluate the state of heart function in clinic. Life is the most complex nonlinear dynamic system, and the heart is the core of the circulatory system, which determines the nonlinearity and complexity of the heart sound signal produced by the heart vibration. In order to simplify and abstract the complex heart system, an ideal linear model has been established, and the linear system is analyzed and processed by time-domain, frequency-domain, time-frequency conversion and so on. However, for half a century, it has been found that linear analysis is not sufficient to study the essentially nonlinear activities of life. As a very important motion form of nonlinear system, chaos can well reveal the special regularity of the inherent randomness of nonlinear process, so this paper intends to analyze the heart sound signal from the point of view of chaos theory. In order to realize the computer-aided diagnosis of heart disease based on heart sound signal, we can deeply understand the inherent characteristics of heart sound signal in essence. In order to improve the recognition accuracy and classification accuracy of heart sound signal, the method of wavelet packet analysis and chaos theory is used to extract and classify the heart sound signal. Compared with the wavelet transform, the wavelet packet has stronger time-frequency resolution, so it can extract the local finer time-frequency information of the original signal. On the one hand, wavelet packet is used to analyze the heart sound signal from time-frequency angle, the heart sound signal is decomposed into different frequency bands by wavelet packet, and then the energy feature of the decomposed frequency band is extracted. In addition, the signal which can represent the characteristics of heart sound signal is decomposed from the component of heart sound signal decomposed by wavelet packet, and the chaotic analysis is carried out, including qualitative and quantitative analysis, in which qualitative analysis includes phase diagram and recursive diagram of heart sound signal. Quantitative analysis includes correlation dimension, maximum Lyapunov exponent and other chaotic characteristic parameters. Then the energy feature of the wavelet packet decomposition is combined with the chaotic characteristic parameter to form the characteristic parameter vector of the heart sound signal, and then the energy characteristics of each frequency band and the chaotic characteristic parameter of the heart sound signal wavelet packet are analyzed by genetic algorithm. The optimal feature vector which can represent the heart sound signal is selected. Finally, the support vector machine (SVM) is used as the classifier and the heart sound signal feature vector is used as the input to realize the automatic classification and recognition of the heart sound signal. The normal and several kinds of abnormal cardiac sound signals, such as premature beat arrhythmia, mitral stenosis, first heart sound division, aortic insufficiency and ventricular septal defect, were detected by the designed heart sound acquisition system. The method described in this paper is used for testing. The results showed that the chaotic qualitative and quantitative characteristics of normal and abnormal heart sounds were significantly different, and the correlation dimension and maximum Lyapunov index of abnormal heart sounds were higher than those of normal heart sounds. It shows that abnormal heart sound signal has high complexity. The combination of wavelet packet energy and chaotic characteristics can obtain a high recognition rate, which shows that chaotic features play an important role in revealing the nonlinear characteristics of heart sound signals. It lays a foundation for the diagnosis of heart sounds and the study of the nonlinear nature of heart sounds.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0
本文编号：2399848
[Abstract]:Heart sound is one of the important physiological signals of human body, which can reflect the mechanical movement of heart and large vessels. It is the most basic noninvasive and convenient method to evaluate the state of heart function in clinic. Life is the most complex nonlinear dynamic system, and the heart is the core of the circulatory system, which determines the nonlinearity and complexity of the heart sound signal produced by the heart vibration. In order to simplify and abstract the complex heart system, an ideal linear model has been established, and the linear system is analyzed and processed by time-domain, frequency-domain, time-frequency conversion and so on. However, for half a century, it has been found that linear analysis is not sufficient to study the essentially nonlinear activities of life. As a very important motion form of nonlinear system, chaos can well reveal the special regularity of the inherent randomness of nonlinear process, so this paper intends to analyze the heart sound signal from the point of view of chaos theory. In order to realize the computer-aided diagnosis of heart disease based on heart sound signal, we can deeply understand the inherent characteristics of heart sound signal in essence. In order to improve the recognition accuracy and classification accuracy of heart sound signal, the method of wavelet packet analysis and chaos theory is used to extract and classify the heart sound signal. Compared with the wavelet transform, the wavelet packet has stronger time-frequency resolution, so it can extract the local finer time-frequency information of the original signal. On the one hand, wavelet packet is used to analyze the heart sound signal from time-frequency angle, the heart sound signal is decomposed into different frequency bands by wavelet packet, and then the energy feature of the decomposed frequency band is extracted. In addition, the signal which can represent the characteristics of heart sound signal is decomposed from the component of heart sound signal decomposed by wavelet packet, and the chaotic analysis is carried out, including qualitative and quantitative analysis, in which qualitative analysis includes phase diagram and recursive diagram of heart sound signal. Quantitative analysis includes correlation dimension, maximum Lyapunov exponent and other chaotic characteristic parameters. Then the energy feature of the wavelet packet decomposition is combined with the chaotic characteristic parameter to form the characteristic parameter vector of the heart sound signal, and then the energy characteristics of each frequency band and the chaotic characteristic parameter of the heart sound signal wavelet packet are analyzed by genetic algorithm. The optimal feature vector which can represent the heart sound signal is selected. Finally, the support vector machine (SVM) is used as the classifier and the heart sound signal feature vector is used as the input to realize the automatic classification and recognition of the heart sound signal. The normal and several kinds of abnormal cardiac sound signals, such as premature beat arrhythmia, mitral stenosis, first heart sound division, aortic insufficiency and ventricular septal defect, were detected by the designed heart sound acquisition system. The method described in this paper is used for testing. The results showed that the chaotic qualitative and quantitative characteristics of normal and abnormal heart sounds were significantly different, and the correlation dimension and maximum Lyapunov index of abnormal heart sounds were higher than those of normal heart sounds. It shows that abnormal heart sound signal has high complexity. The combination of wavelet packet energy and chaotic characteristics can obtain a high recognition rate, which shows that chaotic features play an important role in revealing the nonlinear characteristics of heart sound signals. It lays a foundation for the diagnosis of heart sounds and the study of the nonlinear nature of heart sounds.
【学位授予单位】：重庆大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0
【参考文献】
相关期刊论文 前10条
1 王衍文,王海滨,程敬之;广义时频分析及其在PCG信号分析中的应用[J];北京生物医学工程;1998年02期
2 王兴元,朱伟勇,顾树生;利用关联维数分析心脏的动力学特征[J];东北大学学报;1999年06期
3 朱冰莲;吕佶;;时频表达在心音研究中的应用[J];重庆大学学报(自然科学版);2006年05期
4 张雨;任成龙;;确定重构相空间维数的方法[J];国防科技大学学报;2005年06期
5 杨志安，王光瑞，陈式刚;用等间距分格子法计算互信息函数确定延迟时间[J];计算物理;1995年04期
6 吕振环,吴素文,李喜霞;论混沌学的发展、特性及其意义[J];沈阳农业大学学报(社会科学版);2004年01期
7 胡睿;陈惠;李晓雅;王天鹅;;慢病已成为中国不可承受之重[J];中国社区医师;2011年32期
8 林嘉宇,王跃科,黄芝平,沈振康;语音信号相空间重构中时间延迟的选择──复自相关法[J];信号处理;1999年03期
9 严勇,董筠,楼正国;心音的非线性时间序列分析[J];中国医学物理学杂志;2004年03期
10 王兴元,顾树生;心电动态生理及病理信息的非线性动力学研究[J];中国生物医学工程学报;2000年04期
相关会议论文 前1条
1 丁晓蓉;宋陆州;包鑫;郭兴明;;心音采集及其混沌动力学分析方法的初步研究[A];中国电子学会第十七届信息论学术年会论文集[C];2010年
相关硕士学位论文 前6条
1 董筠;心音的非线性时间序列分析[D];浙江大学;2003年
2 王恒迪;非线性动力学在房颤研究中的应用[D];北京工业大学;2003年
3 周静;心音信号分析方法的研究及其分析系统的开发[D];重庆大学;2004年
4 姜桂仁;混沌时序的特征量分析及相空间重构[D];江苏大学;2005年
5 王丹丹;混沌在心电数据分析中的应用[D];中国海洋大学;2006年
6 吴玉春;心音信号自动识别算法的研究[D];重庆大学;2009年
本文编号：2399848
本文链接：https://www.wllwen.com/yixuelunwen/swyx/2399848.html
