【作者】 何炜琨；

【导师】 吴仁彪；

【作者基本信息】 天津大学 ， 信号与信息处理， 2012， 博士

【摘要】 机场建设迅速发展过程中，机场场道质量问题给民航安全构成严重隐患。论文以基于探地雷达的机场场道监测技术为研究内容，对探地雷达在机场道面结构层中传播的回波信号仿真、机场道面结构层电磁特性反演以及机场道面下的脱空灾害目标的厚度估计等关键技术进行了重点研究，论文主要的研究成果及结论如下：1.研究了探地雷达发射脉冲信号在机场道面有耗介质层中的传输模型及电磁波在道面下介质层中的非恒Q衰减特性，提出了一种基于非恒Q衰减模型的回波仿真方法。给出了电磁波在层状色散介质中的回波信号模型。该方法在运算复杂度上优于基于Maxwell方程的FDTD回波仿真方法，在适用范围上广于基于恒Q衰减模型的回波仿真方法。2.探地雷达在道面结构层中传播的正演模型基础上，研究了基于系统辨识方法的道面结构层电磁特性及其厚度的反演方法。首先在传统的系统辨识方法反演介质层介电常数实部与虚部算法的基础上，考虑了雷达发射带宽内各频率分量对介质层介电常数虚部的影响，提出了基于改进系统辨识方法的介质层介电常数及电导率的反演方法。其次基于超分辨率时延估计WRELAX算法估计反射回波的时延，从而进一步实现了含有薄层道面结构的回波信号检测及其电磁特性的估计。最后针对钢筋加固区介质层电磁特性的反演问题，研究了基于Hyp-curvelet变换的钢筋回波抑制算法，并基于钢筋抑制后的回波数据进行道面介质层电磁特性的反演。3．提出了基于S变换的脱空薄层的检测方法。机场道面下的脱空层厚度很小（一般在1cm以内），对于一般的GPR系统，仅靠时间分辨率是无法将其分开的。S变换采用宽度随频率呈反比变化的高斯窗函数对回波信号进行时频分析，机场场道道面脱空层回波信号所对应的S变换在时频域内彼此混叠，从而产生形状失真，S变换在低频处的频率分辨率较高，因而可以通过检测脱空层回波信号所对应的S变换在低频处的波形失真检测机场场道道面脱空薄层。4．针对探地雷达的无损探测技术，研究了基于复合回波峰值频率上移特性及谱反演优化的脱空薄层厚度估计算法。上述两种脱空薄层厚度估计算法中，均要求获得与脱空薄层上下界面所对应的时域脉冲对。探地雷达系统所接收到的信号是发射脉冲在道面下介质层中传播的回波信号，因而需要将与脱空薄层所对应的时域脉冲对从接收信号中提取出来。本论文利用电磁波在道面介质层结构中的传播模型及改进的系统辨识算法成功提取了与脱空层所对应的时域脉冲对，进而基于峰值频率上移特性及谱反演优化方法实现了脱空薄层的厚度估计。更多还原

【Abstract】 With the rapid development of runways construction, the quality problem ofrunways is a serious hidden trouble for the safety of aviation. In the dissertation, thekey techniques concerned in the field of runways surveillance based on GroundPenetrating Radar (GPR) are studid. The main contents consist of the echo simulationof electromagnetic (EM) wave propagated in the pavement structure of runways, theEM properties inversion of the pavement structure and the thickness calculation ofvoid disaster in runways.The main achievements and conclusion of this paper are summarized asfollows:1. The propagation model and non-constant–Q attenuation model of the transmittedEM wave in the loss multilayer media are studied. A GPR echo simulation methodbased on non-constant-Q attenuation model is proposed.and the echo signal modelof EM wave propagated in the dispersive loss media layers is constructed. Theproposed echo simulation method is better than FDTD method based on Maxweillfunction in calculation burden, and is wider than that based on constant-Qattenuation model in the applied area.2. Based on the propagation model of EM wave propagated in the pavementstructiure, the EM properties and depth inversion method based on systemidentification theory is also struded. Firsty, based on the conventional systemidentification method to be used to estimate real part and imaginary part of thecomplex permittivity, a new permittivity and conductivity inversion method basedon the improved system identification is proposed, in which the effect of thefrequency proponents within the transmitted band width on the imaginary part ofcomplex permittivity is considered. Secondly, The EM property of multilayermedia containing the thin layer are intensively given, in which a super resolutionparameter estimation method,WRELAX, is introduced to estimate the time delayof echo. Finally, as far as the EM property inversion of the reinforcing steel bararea in runways is concerned, the bar suppression algorithm based onHyp-cruvelet transform is studied, and then the EM property are estimated basedon the bar suppressed data. 3. The detection method of the void disaster based on the time-frequency distributioncharacteristic of S transform is proposed. The void layer is thin layer, whosethickness is less than1cm, and so for the current GPR system, the void thin layercan not be detected only by the time resolution. S transform is a time-frrequencytool in whch the Gaussian window width is inverse proportion to the frequency.The S transform of the echoes reflected from the the void layer are aliased andeffected each other. S transform has higher frequency resolution in the lowerfrequency, where the variance of the S transform can be detected. Therefore thevoid layer can be detected by the distortion of the bottom part of S transformobtained by the reflected echoes.4. For the non-destructive detection characteristic of GPR, the void thicknessestimation algorithms based on the upper-shifted peak frequency property of theechoes and spectrum optimization inversion are studied. For the above twomethods, the impulse pair in the time domain corresponding to the upper andbottom interface of void-layer are needed. For the non-destructive GPR system,the received signal is the reflected echo of the EM wave propagated in thepavement structure (including the void layer). So the impulse pair in the timedomain corresponding to the upper and bottom interface of void-layer are neededto be extracted. In this dissertation, they are extracted based on the EM wavepropagation model and system identification method studied in the forgoingchapers, then the void layer thickness is calculated based on the upper-shifted peakfrequency property and spectrum optimization inversion algorithm.更多还原