【作者】 李邦宇；

【导师】 许会；

【作者基本信息】 沈阳工业大学 ， 测试计量技术及仪器， 2012， 博士

【摘要】 生物雷达是用来探测未知区域中生命体地理信息和生命信息的一种仪器，其应用领域包括灾害救援、医疗监护、城市安保与反恐等。生命体的地理信息是指生命体在未知区域中的空间位置；生命体的生命信息是指呼吸、心跳等生命信息。生物雷达通过计算机层析技术获得未知区域的水平断层图像，在图像中可以确定生命体的地理信息，应用时往往需要穿透障碍物，障碍物在雷达回波信号中表现为很强的直达波，影响图像中对目标的分辨。采用雷达检测人体的呼吸、心跳等微动信息，常利用微多普勒效应，其难点在于在强噪声中对微弱信号的检测。已有的研究成果不能完全解决直达波去除和呼吸、心跳信息检测问题，应用具有一定局限性。因此，在生物雷达回波信号的处理和系统参数方面做深入的研究，对提高其性能具有重要意义。本文针对以上问题开展研究，主要内容如下：（1）超宽带步进变频连续波雷达近场成像算法研究。首先介绍了近场成像的概念，针对超宽带步进变频连续波生物雷达，建立了其正问题数学模型。然后对生物雷达成像中的合成孔径技术进行了介绍，并研究一种应用于超宽带步进变频连续波雷达的频率-波数域成像算法原理，通过仿真分析了该算法的成像性能。（2）生物雷达直达波抑制方法研究。首先理论分析了直达波对于生物雷达成像的影响，并通过仿真进行了验证。针对传统去除直达波信号方法的不足或者局限性，本文提出了基于奇异值分解和复数FastICA两种去除直达波的方法。奇异值分解方法利用回波数据中直达波分量占主要成分的特性，在一定程度上克服了传统对消法在去除直达波的同时也削弱了目标信号幅度的缺点，提高了处理后的信噪比，并且算法快速；复数FastICA方法是从回波信号的数据结构入手，利用了直达波信号与目标信号的高阶统计量信息，把信号分解为多个相互独立的信号分量，使各个分量之间非高斯性最强，此方法运算量较大，但是直达波抑制效果较好。最后，通过实验对这两种方法进行了验证，并提出了目标信号/噪声能量比的定义，通过目标信号/噪声能量比和信干比两项指标将对消法、奇异值分解方法和复数FastICA方法进行了对比。（3）生命体微多普勒信息检测方法研究。针对传统采用单频连续波发射信号检测生命体微多普勒信息方法抑制噪声能力较弱，不能检测多个目标的微多普勒信息，以及不能在检测的同时对目标定位的不足，本文提出了采用超宽带步进变频连续波信号的生物雷达，并提出了一种距离域滤波的方法检测生命体为多普勒信息。首先针对超宽带步进变频连续波生物雷达检测生命体呼吸信号进行了数学建模，并针对不同呼吸频率时目标距离域一维成像进行仿真，通过对仿真结果的分析，得到在呼吸频率为1Hz以下，呼吸信号几乎附着在目标距离域一维像当中，并且做了距离域一维像截取的仿真实验，得到在距离域一维像两边一定范围内做截取不会丢失呼吸信号的结论。根据以上结论，本文提出了通过距离域滤波方法来消除噪声和干扰，实现检测生命体呼吸频率的目的。通过实验对上述方法进行了验证。单人试验中，呼吸频率能很好的被检测出来；双人实验中，由于多径干扰的存在，使得检测误差增大，甚至会检测不到呼吸频率。针对背景噪声较强，影响对目标距离域一维像的判断这种情况，本文采用距离域峰值跟踪方法，在强背景噪声的情况下对生命体在距离向进行定位并检测出其呼吸频率。（4）测量系统参数与呼吸频率检测中采样时间累计长度的研究。在测量系统中，天线参数各频点的不一致性、发射信号的步进频率间隔以及接收端中频滤波器带宽的设置对于测量的影响较大，本文通过对测量环境的分析以及理论推导，提出了相量差法去除天线的不一致性，以及步进频率间隔和中频滤波器带宽的合理设置。之后本文分析了采样时间累积长度对呼吸频率检测的影响，结合前面的实验结果，提出了合理的采样时间累积长度。在测量系统参数和采样时间累积长度确定后，本文对基于距离域滤波方法检测呼吸频率的误差和分辨力进行了分析。更多还原

【Abstract】 Biological radar is an instrument which is used to detect the geographic informationand life information of life body in an unknown area. Its applications include disaster relief,medical application, security application and fight against terrorism and other occasions.The geographic information of life body is the spatial location of life body in an unknownarea. The life information of life body is respiratory rate, heartbeat rate and otherparameters.Biological radar obtain horizontal tomography image of the unknown region throughcomputer tomography, and the geographic information of life body can be determined inimage. Biological radar often needs to penetrate obstruction. The obstacle is manifested asa strong direct wave in the echo signal of radar, which affect to distinguish the target in theimage. Micro-Doppler effect is often used when using radar to detect human breathing,heartbeat and other micro-information. The difficulty is that weak signal is needed to bedetected in strong noise. Existing research results can not completely solve the problem ofdirect wave removal and the detection of breathing rate and heartbeat rate, so theapplication of it has certain limitation. Therefore, in-depth study in the echo signalprocessing and system parameters of biological is great significance to improve itsperformance. Doing research for the above problem in this paper, the main content is asfollows:（1）The research for near-field imaging algorithm of ultra-wideband steppedfrequency continuous radar. Firstly, the concept of near-field imaging is introduced. Themathematical model of the forward problem is established for ultra-wideband steppedfrequency continuous wave biological radar. Then synthetic aperture technology in biological radar imaging is introduced, The application of frequency-wavenumber domainimaging algorithm used in ultra-wideband stepped frequency continuous radar is studied,and the imaging performance of the algorithm is analyzed by simulation.（2）The research for direct wave suppression method of biological radar. Firstly, theeffect of direct wave to the imaging of biological radar is analyzed by theory. Fordeficiencies or limitations of traditional method used in the removal of direct signal, thesingular value decomposition method and complex FastICA method for removing directwave is proposed in this paper. The characteristic that the direct wave component accountsfor the main component in echo data is used in the singular value decomposition method.This method overcomes the shortcoming of the traditional Interference Canceling methodthat the amplitude of target signal is weakened after removing the direct signal. The signalto noise ratio of image is improved, and algorithm is fast. The complex FastICA method isproceed from the data structure of the echo signal.It is based on the higher order statisticscharacteristic of signal, and can decompose the signal into multiple independent signalcomponents, makes the strongest non-Gaussian among the various components of thesignal. Finally, the feasibility and effectiveness of the singular value decompositionmethod and the complex FastICA method is veritied in the experiment. The definition ofsignal to interference ratio of energy is proposed in this paper, and the comparement isdone among of Interference Canceling method, singular value decomposition method andcomplex FastICA method using two evaluations of signal to interference ratio of energyand signal to interference ratio.（3）The research for micro-Doppler information detection method of life body. Forthe deficiencies of detecting life body micro-Doppler information methods usingtraditional single-frequency continuous-wave emission signal, which is that the noisesuppression capability is weak, micro-Doppler information of multiple targets can not bedetected and the location of target can not be obtained when detecting micro-Dopplerinformation, ultra-wideband stepped frequency continuous wave emission signals anddistance domain filtering method is proposed in this paper for detecting life body micro-Doppler information. Firstly, the mathematical modeling of life body breathingsignal detection for ultra-wideband stepped frequency continuous wave biological radar isdone. Simulation is done for one-dimensional imaging of target in distance domain ofdifferent respiratory rate. Through the analysis of simulation results, the respiratory signalis almost attached to one-dimensional imaging of distance domain when the respiratoryrate is less than 1Hz. The simulation of interception in the one-dimensional imaging ofdistance domain is done, and the conclusion is obtained that the respiratory signal is notlost when doing interception at a certain distance of both sides for one-dimensionalimaging of distance domain. Based on the above conclusions, a distance domain filteringmethod is proposed in this paper for eliminating the noise and interference, and achievingthe detection purpose of breathing frequency of life body. The method described above isverified by experiment. In single body experiment, the respiratory rate can be detected. Indouble body experiment, the detection error increases due to the presence of multipathinterference, and even the respiratory rate detection fails. For strong background noise,which affects the judgement for one-dimensional distance domain imaging of target, adistance domain peak tracking method is used in this paper. This method can locate the lifebody and detect its respiratory rate in the case of strong background noise.（4）The research for measurement system parameters and sampling time cumulativelength of respiratory rate detection. In the measurement system, the parametersinconsistency of each frequency point of the antenna, the stepping frequency interval of thetransmitted signal and the IF filter bandwidth settings of the receiver affect measurementgreatly. In this paper, through the analysis of measurement environment and theoreticalderivation, the phasor difference method is proposed for removing the inconsistency ofantenna, and reasonable set of step frequency interval and the IF filter bandwidth.Then thispaper analyzes the effect of sampling time cumulative length for the detection ofrespiratory rate.Combined with the previous experimental results, a reasonable samplingtime cumulative length is proposed. Finally, the error and resolution of the respiratory rate detection based on the distance domain filtering method are analyzed after themeasurement system parameters and the sampling time cumulative length is determined.更多还原