【作者】 潘水洋；

【导师】 冯久超；

【作者基本信息】 华南理工大学 ， 信号与信息处理， 2011， 硕士

【摘要】 穿墙生命探测雷达(Through-the-wall Detection Radar, TWDR)是灾后搜寻幸存者的重要工具,它发射电磁波穿过墙壁,碰到人体返回,通过对回波进行处理,就可以获取目标的呼吸、心跳等生命信号。生命信号识别是穿墙生命探测雷达研究的核心问题。生命信号是一种受强噪声干扰的微弱信号,信号随机性强,具有非平稳特性,给呼吸心跳信号的分离、多目标的识别带来很大困难。本文在研究连续波穿墙生命探测雷达(Continuous Wave TWDR,CW-TWDR)和脉冲超宽带穿墙生命探测雷达(Ultra-Wideband TWDR,UWB-TWDR)原理的基础上,将高分辨率谱估计、自适应信号处理、盲源分离算法、模式识别等现代信号处理技术运用于生命信号的识别。先进的信号处理技术不仅提高了生命探测雷达生命信号的识别精度,而且能够完成多人识别以及人和动物区分。本文开展了如下三方面的工作：(1)针对冲激超宽带生命探测雷达中,传统的平均相消法不能有效去除直达波,从而影响目标回波信号的提取这一问题,采用宽带互相关法对回波时延进行估计,然后提取各个回波时延序列的均值与方差作为特征量,运用C-均值聚类算法对回波进行分类,实现直达波与目标回波的分离；(2)针对穿墙生命探测雷达系统中传统的基于数字滤波和谱估计的生命信号提取方法不能有效处理非平稳信号、易受呼吸谐波干扰的问题,提出了一种基于经验模态分解(Empirical Mode Decomposition,EMD)的生命信号提取的新方法,实验结果表明,新提出的方法能避免呼吸信号谐波对心跳信号的干扰,因而能更加精确的提取呼吸和心跳参数；(3)最后基于Labview构建了雷达硬件系统测试平台,结合仿真和实测两个方面加以验证,逐步完善整套信号处理算法,为实现更经济实用的雷达生命探测仪,提供了理论支持。更多还原

【Abstract】 Through-the-wall Detection Radar (TWDR) is widely used in rescuing human victims buried after earthquake. The radar emits a microwave beam to penetrate the wall and reach the subject, the reflected wave from the human body is received and heartbeat and respiration signals are extracted. One challenge is how to get the breathing and heartbeat information from the echo signal because vital signals are weak, time-varying and non-stationary signals.In this thesis, Continuous Wave TWDR and Impulse-based UWB TWDR have been applied to detect vital signals. Advanced signal processing methods, such as high-resolution spectral estimation, adaptive signal processing, blind separation algorithm, pattern recognition, are used for vital signals detection. These have the potential to not only improve the robustness of Doppler radar sensing to practical levels, but to also make possible the gathering of additional information, such as determining the number of subjects in a particular environment and recognise people and animals.The work of this thesis is reflected in the following areas:Firstly, a new method to suppress the direct wave in impulse-based UWB radar system is proposed. Wide-band cross correlation methods is used to estimate the echo delay. Then, select mean and variance of the delay sequence as features and use C-Means clustering algorithm to achieve echo classification, so direct wave is removed effectively. The results indicate that the proposed method more effective than mean-subtraction method to suppress the direct wave. Secondly, in order to overcome the shortcomings that the conventional vital signal extraction algorithms based on FFT are not good at dealing with non-stationary signals and susceptible to harmonic interference, a new method for vital signal extraction based on empirical mode decomposition is proposed in this paper. The proposed algorithm greatly improves the estimation accuracy of respiratory rate and heart-rate. Lastly, we establish radar hardware test platform based on Labview. Combining simulation and measured data to verification and improve the algorithms, which provide theoretical support for achieving practical radar detector.更多还原