【作者】 周红标；

【导师】 陈若珠；

【作者基本信息】 兰州理工大学 ， 控制理论与控制工程， 2009， 硕士

【摘要】 心血管疾病是当今危害人类健康的主要疾病之一,随着我国人民生活水平的不断提高和人口的逐渐老龄化,其发病率和死亡率日趋增加。因此,需要一种无创检测方法来随时了解自身的健康状况。脉搏信号是一种非线性、非平稳的微弱生理信号,含有大量的生理、病理信息,常对其进行检测分析,可以达到对心血管疾病的早预防和早治疗的目的。目前脉率大都被静态检测,如能动态地连续记录人体脉率,必然会揭示人体血管病变的某些规律。本文的主要工作就是设计并实现一套动态脉率数据采集与脉搏信号处理系统,并对正常人群和心血管疾病患者脉搏信号特征进行分析。首先,本文设计了脉率数据与脉搏信号采集系统,确定以AVR系列ATmegal6单片机为核心,对脉率数据与脉搏信号进行采集和存储,并通过USB接口将数据传输到上位机。在上位机以虚拟仪器软件LabVIEW为平台,采用多面板方式和模块化设计思想开发一套脉搏信号分析系统软件。该软件包括患者管理模块、数据采集模块、信号处理模块和数据远程传输模块等。其次,由于脉搏信号信噪比较低,在采集过程中混入的肌电干扰、工频干扰、基线漂移等噪声信号,对脉搏信号特征提取容易造成误判,需要进行去噪处理。本文在分析D.L.Donoho提出的小波硬阈值和软阈值方法特点的基础上,研究了一种新的小波自适应阈值消噪方法,仿真试验证明了该方法优于传统的阈值消噪法。最后,采集学生和临床确诊为心血管疾病患者的脉搏信号,利用小波分析方法提取脉搏信号各尺度能量值,针对脉搏信号频带特点,改进了特征提取算法。通过统计分析,在有限样本的情况下,给出了区分正常人群和心血管疾病患者的特征值范围,实验数据表明该方法取得了很好的识别效果。由于小波包变换具有任意多尺度分解特性,它是一种比小波变换更加精细的分析方法,因此本文进一步研究了脉搏信号小波包分析法,该方法正确识别了被小波算法漏检的病脉信号。更多还原

【Abstract】 Cardiovascular diseases are the major factor that harms the human health.As the rapid progress in people’s living condition and the greying of Chinese population,the morbidity and mortality of cardiovascular diseases increase gradually,people need a non-invasive method to examine their health status.The pulse signal is a kind of non-linear,non-stationary as well as weak signal,which contains abundant human physiological and pathological information,plays a significant role in making early diagnosis and treatments for cardiovascular diseases.The pulse rate is detected in static states now.If can record in long-term continuously,it will reflects changes in blood vessel systems of human body.The work in this paper is mainly about the design and realization of dynamic pulse rate data and pulse signal processing.To begin with,the acquisition system of pulse rate data and pulse signal is designed by MCU of AVR ATmega16.Both the cost and the quality are also considered. The hardware device can acquire and store data,and communicate with PC using USB interface.The virtual pulse signal processing system is designed in PC based on vitual instrument software LabVIEW,includes the management of patient information、the acquisition of data、the process of signal、the communication of data,which adopts the approach of Multi-panel and the idea of modular design.Secondly,the SNR of pulse signal is lower than others.In order to avoid the wrong identification from the characteristics of pulse signal,to remove the EMG interference, powerline interference and baseline-drift existing in the course of collecting pulse signal data were proved importantly.Based on Donoho’s method of wavelet transformation (WT),an improved wavelet denoising method with adaptive thresholding is given,which has both advantages of hard-thresholding and soft-thresholding.The simulation experiment indicates that the proposed method is better than traditional wavelet thresholding denoising methods.Finally,the pulse signals of students and patients of crdiovascular diseases are analyzed.The energy of every scale is extracted based on WT.Aiming at the characteristics of pulse signal frequency band,the algorithm of feature extraction is improved.The experiment datum indicate that the range of features can distinguish pulse signals of patients from limited samples,which presents a good recognition accuracy. Because wavelet packet transform(WPT) can provide an arbitrary time-frequency decomposition for the signals,which is a more refined method of signal processing. Further,application in feature extracting of pulse signal based on WPT is put forward,it can identify the signal of patient which is recognised incorrectly by WT.更多还原