心冲击信号的经验模态分解提取方法研究

发布时间：2018-02-26 01:05

本文关键词： 睡眠监测经验模态分解心冲击信号边界效应心率变异性　出处：《哈尔滨工业大学》2017年硕士论文　论文类型：学位论文

【摘要】：夜间的睡眠质量影响着人们的日常生活和身体健康。良好的睡眠是维持身心健康的重要保证。在睡眠过程中,正常平稳的呼吸和心跳能够促进血液循环,为身体提供充足的氧气支持新陈代谢,提高睡眠质量。高质量的睡眠能够增强自身免疫力,延长细胞的活性,是青少年人身心健康成长必不可少的条件。同时,中老年人保持充足健康的睡眠,可以使身体和头脑在夜间得到充分的休息,为日常的生活和工作提供健康保障。如今,社会的快速发展使人们的生活方式发生转变,为了有充沛的精力应对每天的生活和工作,人们对于睡眠质量也有了更多的关注和更高的要求。另外,由于科技不断发展,生活成本显著增高,人们的日常需求不断增加和升级,这也为自己和家人带来了更多的压力,容易引发失眠、睡眠呼吸暂停等睡眠障碍问题。课题组针对中老年人的睡眠健康问题,设计出睡眠监测床垫实时监控睡眠状态,综合人体体征指标分析身体状况。本课题利用硬件系统采集到的人体体征信号进行信号处理相关技术研究,选取经验模态分解信号处理方法提取分析人体心冲击信号。经验模态分解是一种新型的信号处理技术,其良好的自适应性非常适用于分析非线性、非平稳信号。基于传统的经验模态分解信号处理技术,本论文为了更好地适应所研究的信号特征对算法进行了改进,以期望达到最佳分解效果。应用改进后的经验模态分解算法实现对信号的分解提取,针对处理后的心冲击信号进行了心率变异性分析。心率变异性表示人体正常心跳周期间的微小涨落,蕴含着大量人体心血管疾病的信息,利用心冲击信号分析得出心率变异性的相关指标,能够及早判断和发现心血管疾病的发生,实现日常睡眠的监护。本论文阐述了经验模态分解算法的基本原理,介绍了瞬时频率和特征模态分量的基本概念,给出了详细的分解步骤以及希尔伯特变换的定义式,并且应用添加高斯白噪声的方法解决了在分解过程中的模态混叠问题。在算法改进方面,一方面对包络拟合方法进行了优化。另一方面,分析了最相关波形拟合法等边界效应处理方法,最终采用上下包络边界极值点延拓法解决了边界飞翼问题。而后,本文采集了8组正常人的试验数据,利用改进后的经验模态分解算法对其进行分解提取出心冲击信号。最后,利用处理得到的信号综合时域、频域、非线性三种分析方法进行心率变异性分析,对比指标参数,验证性能。
[Abstract]:The quality of sleep at night affects people's daily life and health. Good sleep is an important guarantee of maintaining physical and mental health. During sleep, normal and steady breathing and heartbeat can promote blood circulation. Provide adequate oxygen to the body to support metabolism, improve sleep quality. High quality sleep can enhance their own immunity, prolong the activity of cells, is an essential condition for the physical and mental growth of teenagers. Adequate and healthy sleep for middle-aged and old people allows the body and mind to rest at night and provide health protection for daily life and work. Nowadays, the rapid development of society has transformed people's way of life. In order to have a lot of energy to cope with daily life and work, people also have more attention to sleep quality and higher requirements. In addition, because of the continuous development of technology, the cost of living has increased significantly. People's daily needs are increasing and upgrading, which also brings more pressure to themselves and their families, which can easily lead to insomnia, sleep apnea and other sleep disorders. The sleep monitoring mattress is designed to monitor the sleep state in real time, and the body condition is analyzed by synthesizing the physical sign index. The signal processing related technology is studied by using the human sign signal collected by the hardware system. The empirical mode decomposition (EMD) signal processing method is selected to extract and analyze the human heart shock signal. EMD is a new signal processing technology, and its good adaptability is very suitable for nonlinear analysis. Non-stationary signal. Based on the traditional empirical mode decomposition signal processing technology, this paper improves the algorithm to better adapt to the studied signal characteristics. In order to achieve the best decomposition effect, the improved empirical mode decomposition algorithm is used to extract the signal. Heart rate variability (HRV) is used to analyze the heart rate variability (HRV), which represents the small fluctuation during the normal heartbeat, and contains a large amount of information about cardiovascular diseases. The related indexes of heart rate variability can be obtained by the analysis of cardiac shock signal, and the occurrence of cardiovascular disease can be judged and discovered early, and the monitoring of daily sleep can be realized. In this paper, the basic principle of empirical mode decomposition algorithm is described. The basic concepts of instantaneous frequency and characteristic modal component are introduced, and the detailed decomposition steps and the definition of Hilbert transform are given. And the method of adding Gao Si white noise is used to solve the problem of modal aliasing in the process of decomposition. In the aspect of algorithm improvement, the envelope fitting method is optimized on the one hand, and on the other hand, The boundary effect processing methods such as the most correlated waveform fitting method are analyzed, and the upper and lower envelope boundary extremum continuation method is used to solve the boundary flight wing problem. Then, the test data of 8 groups of normal people are collected. The improved empirical mode decomposition algorithm is used to extract the heart-impact signal. Finally, the heart rate variability is analyzed by the three methods of integrated time-domain, frequency-domain and nonlinear analysis, and the index parameters are compared. Verify performance.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R740;TN911.7

【参考文献】