面向无线躯感网的ECG信号处理算法的研究与实现
文献类型：学位论文
| 作者 | 凤超 |
| 学位类别 | 硕士 |
| 答辩日期 | 2012-05-28 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 中国科学院沈阳自动化研究所 |
| 导师 | 梁炜 |
| 关键词 | 无线躯感网 HMM 信号滤波 特征提取 信号分类 |
| 其他题名 | Research and Implementation of ECG Signal Processing Algorithm Oriented to Wireless Body Sensor Network |
| 学位专业 | 模式识别与智能系统 |
| 中文摘要 | 心血管疾病已经成为导致人类死亡的最主要疾病之一。心电图（Electrocardiogram, ECG）是检测、评估和诊断心血管疾病的主要手段。心脏病“一过性”、“瞬时性”等特点使得传统的静态和动态ECG检测仪器无法及时检测到紧张劳累、精神受到刺激、情绪激动等特殊状态下的异常心电信息。面向无线躯感网的ECG信号检测技术能够实现连续、长时间的心电信息检测，在灵活性、高效性、时效性和经济性等方面具有明显优势，已成为动态ECG的发展趋势，具有广阔的应用前景。 本文在对ECG信号特点、无线躯感网技术发展以及ECG信号处理相关算法研究现状进行详细总结的基础上，从ECG信号预处理、ECG信号特征提取、ECG信号分类三个方面对ECG信号处理算法进行了研究，取得以下成果： 1. 针对无线躯感网资源受限、以及其移动性、电极接触不良等所导致的ECG信号易受50Hz工频和低频基线漂移信号干扰等问题，本文提出了基于简单整系数带阻滤波器的ECG信号预处理算法，仿真结果显示该算法不仅能有效地滤除工频干扰信号和基线漂移信号，而且能滤除高频的谐波干扰，同时该算法复杂度低，适合无线躯感网应用。 2. 针对ECG信号非线性、随机性、非平稳性等特点以及无线躯感网的算法实时性要求，提出了基于隐马尔科夫模型（Hidden Markov Model, HMM）的ECG信号特征提取算法，建立了面向ECG信号特征提取的HMM模型，并优化了模型训练过程中的初始值。仿真结果显示本算法能准确地提取ECG信号的特征信息，同时所提算法复杂度低，能够满足无线躯感网对算法实时性的要求。 3. 针对ECG信号微弱且混沌导致的诊断困难问题，在ECG信号特征提取的基础上，提出了一种基于HMM的低开销、快速ECG信号分类算法，建立了用于分类心室早期收缩（Premature Ventricular Contraction，PVC）和心房早期收缩（Atrial Premature Contraction，APC）的HMM模型，设计了相应的模型训练方法。仿真结果表明该算法能从正常信号中有效分类PVC信号和APC信号，同时算法能够满足无线躯感网对算法的低复杂度要求。 为了验证所提面向无线躯感网的ECG信号处理算法性能，本文设计和实现了一个面向无线躯感网的ECG信号处理实验平台，实验证明该平台可以有效地支持ECG信号采集、预处理、特征提取、分类等方法和理论的研究。 |
| 索取号 | TN911.7/F68/2012 |
| 英文摘要 | Cardiovascular disease has become one of the main reasons for human death. Electrocardiogram (ECG) is the primary means of detection, assesement, and diagnosis of the cardiovascular disease. Cardiovascular disease is transient and instantaneous, and the traditional static or dynamic ECG detectors can not timely detect the abnormal ECG information under special conditions, such as in nervous and fatigue state, mental simulation state, or emotional state. ECG signal detection technologies based on wireless body sensor network, which have become the trend of dynamic ECG and have broad application prospect, can continuously detect the ECG information for a long time and have many advantages, such as flexibility, high efficiency, timeliness, and economy. The dissertation firstly summaries the features of the ECG signals, the development of wireless body sensor networks, and the related algorithms for ECG signal processing. Then, the dissertation studies the ECG signal processing problem in aspects of signal preprocessing, feature extraction, and signal classification. The innovative achievements are described below. 1. The resources of wireless body sensor network are limited. In addition, because of the mobility and the loose electrode contact, the ECG signal is susceptible to 50 Hz power interference and susceptible to low frequency baseline wander. In order to solve these problems, an ECG signal preprocessing algorithm based on the simple integral coefficient stop filter is proposed. Simulation results show that this algorithm can not only effectively filter the power interference signal and the baseline wander signal, but also filter the high frequency harmonic interference. Moreover, this algorithm has low complexity, which is suitable for applications of wireless body sensor network. 2. The ECG signal has the nonlinear, random, non-stationary features. Moreover, the wireless body sensor network has critical real-time requirement to algorithms. In order to solve these problems, a feature extraction algorithm for ECG signals based on Hidden Markov Model (HMM) is put foward. A HMM model for ECG signal feature extraction is established and the initial parameter values of training are optimized . Simulation results show that this algorithm can correctly extract the ECG signal features, and the complexity is low, which can satisfy the real-time requirement of wireless body sensor network. 3. The ECG signal is so weak and chaotic that diagnosis of Cardiovascular disease is difficult. In order to solve this problem, a low-cost and rapid algorithm for classifying the ECG signals is presented based on the HMM theory and the above extraction results of the ECG signal features. This algorithm establishes a HMM model for classifying the Premature Ventricular Contraction (PVC) and the Atrial Premature Contraction (APC), and designs the corresponding model training method. Simulation results show that this algorithm can effectively classify the PVC signals and APC signals from normal signals, and can satisfy the low complexity requirement of wireless body sensor network. 4. An experimental ECG signal processing platform for wireless body sensor network is designed and developed to validate the performance of the proposed algorithms. The experiment results indicate that this platform can effectively support the method and theory research on ECG signal acquisition, preprocessing, feature extraction, and classification. |
| 语种 | 中文 |
| 公开日期 | 2012-07-27 |
| 产权排序 | 1 |
| 分类号 | TN911.7 |
| 源URL | [http://ir.sia.ac.cn/handle/173321/9338]  |
| 专题 | 沈阳自动化研究所_工业控制网络与系统研究室 |
| 推荐引用方式 GB/T 7714 | 凤超. 面向无线躯感网的ECG信号处理算法的研究与实现[D]. 中国科学院沈阳自动化研究所. 中国科学院沈阳自动化研究所. 2012. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。