起搏心电信号的压缩感知重构方法研究

发布时间：2018-05-11 23:01

本文选题：起搏心电信号 + 压缩感知　；参考：《重庆理工大学》2017年硕士论文

【摘要】：传统起搏器监测方法受制于信号频率、功耗、监测方式等因素,因而在进行实时监测时难以应对大量起搏心电数据的采样、处理与传输。压缩感知理论可以进行信号的低功耗采样,实现监测终端的低功耗、高效率信号获取,并在监测终端利用相应优化算法恢复出逼近原始信号的重构信号。本研究致力于探索一种适用于起搏心电信号的信号感知以及重构方法,为以后利用压缩感知技术实现起搏心电信号远程监测系统的设计提供技术支撑。本课题主要结合了压缩感知理论并针对起搏心电信号自身特性,对其中的信号采样、信号重构等技术进行理论及实验研究。主要从起搏心电的压缩感知过程及重构算法着手,重点针对噪声环境下真实信号的不可知性,将有偏重构的思想引入到重构算法中,并通过实验测试以及理论证明,对提出的算法进行多角度评测,验证了算法的有效性。主要的研究工作有:第一,根据起搏心电信号特点,设计随机稀疏的二值测量矩阵,并应用于后续的信号感知以及重构过程。第二,基于分块稀疏贝叶斯学习,对起搏心电信号的压缩感知进行研究,探索该算法框架下的信号重构。第三,针对噪声环境下的信号重构问题,根据现有压缩感知重构算法的特点,结合岭回归有偏重构思想改进出适用于起搏心电信号的重构算法,在重构精度及重构效率两个方面相较于原有算法均有提升;并对岭参数的优化方法进行研究以确定出噪声环境下起搏心电信号重构的最优岭参数值。第四,对研究内容中测量矩阵的设计效果及重构算法的改进成效进行评价:对测量矩阵从感知效率、矩阵构造等角度进行对比;对重构算法从重构精度及重构效率以及算法对信号、噪声类型的普适性等方面分别对比。
[Abstract]:Traditional pacemaker monitoring methods are limited by signal frequency, power consumption, monitoring mode and other factors, so it is difficult to deal with the sampling, processing and transmission of a large number of pacemaker data in real-time monitoring. The compression sensing theory can sample the signal with low power consumption and achieve the low power consumption and high efficiency signal acquisition. The reconstruction signal approaching the original signal can be recovered by the corresponding optimization algorithm in the monitoring terminal. This study is devoted to exploring a method of signal sensing and reconstruction suitable for pacing ECG signals, which provides technical support for the design of remote monitoring system for pacing ECG signals by using compression sensing technology in the future. This paper mainly combines the theory of compression sensing and the characteristics of pacemaker ECG signals, and carries out theoretical and experimental research on the signal sampling and signal reconstruction techniques. This paper mainly starts with the compression process and reconstruction algorithm of pacing electrocardiogram, focusing on the unknowability of real signal in noise environment. The idea of biased reconstruction is introduced into the reconstruction algorithm, and it is proved by experiment and theory. The effectiveness of the proposed algorithm is verified by multi-angle evaluation. The main research works are as follows: first, according to the characteristics of pacing ECG, a random sparse binary measurement matrix is designed and applied to the subsequent signal sensing and reconstruction process. Secondly, based on block sparse Bayesian learning, the compression perception of pacemaker ECG signal is studied, and the signal reconstruction under the framework of this algorithm is explored. Thirdly, aiming at the problem of signal reconstruction in noisy environment, according to the characteristics of existing compression sensing reconstruction algorithms and combining with the biased conception of ridge regression, the reconstruction algorithm suitable for pacing ECG signals is improved. Compared with the original algorithm, the reconstruction accuracy and efficiency are improved, and the optimization method of ridge parameters is studied to determine the optimal ridge parameters of ECG reconstruction in noisy environment. Fourthly, the design effect of the measurement matrix and the improvement of the reconstruction algorithm are evaluated. The measurement matrix is compared from the aspects of perception efficiency and matrix construction. The reconstruction algorithm is compared in terms of reconstruction precision, reconstruction efficiency and the universality of the signal and noise types.
【学位授予单位】：重庆理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R540.4;TN911.7

【参考文献】