【作者】 戚秀真；

【导师】 赵祥模；

【作者基本信息】 长安大学 ， 智能交通与信息系统工程， 2016， 博士

【摘要】 由于混凝土材料本身的复杂特性以及在设计、施工、后期保养阶段的不足,很多的混凝土构件在建造以及使用过程中,会出现一些缺陷或者是损伤,严重影响了混凝土结构物的使用安全,甚至会造成严重的恶性事故。如何能够快速精确的获得混凝土构件内部的质量情况,为评估混凝土结构物的性能提供准确的依据,显得尤为重要。近年来,在混凝土的无损检测中引入了超声层析成像技术,能够将混凝土结构检测面的质量情况、缺陷大小分布,以图像的形式精确、形象、完整的显示出来,弥补了传统混凝土检测方法的一些局限,成为混凝土结构检测的发展方向。本文从混凝土超声检测基本原理、基于射线理论的混凝土超声层析成像理论及正演反演方法、基于菲涅耳体的混凝土超声层析成像方法、层析成像结果图像后处理方法及影响层析成像结果的主要因素等几个方面,进行系统深入的研究,主要研究成果如下:(1)针对混凝土测区内部波速随机分布、缺陷具有一定规律的特点,提出了一种基于Snell定律与扰动法相结合的改进射线追踪方法,以适用于混凝土超声层析成像的正演过程特点。通过对不同模型的试验结果表明,改进后的射线追踪路径更加接近真实声波路径,基于Snell定律改进的扰动法射线追踪适用于混凝土超声层析成像正演过程。(2)通过对常规混凝土超声层析成像迭代反演算法及最小二乘算法的研究,提出了适用于混凝土超声层析成像的两种改进ART算法、改进的共轭梯度算法及改进的LSQR算法。在两种ART改进算法中分别引入波速分级和成像单元逐级分块的概念;改进共轭梯度算法中引入预处理矩阵对投影矩阵进行预处理,降低投影矩阵的条件数;改进LSQR算法中通过加入阻尼因子来抑制误差造成的影响。通过数值模型及混凝土试块试验,对改进算法与原始算法进行了对比分析,结果表明,改进算法在计算速度和计算精度上都得到了提高。通过大、小两组混凝土试块内部缺陷成像结果试验及分析,验证了改进算法的有效性,成像结果图像表明,改进算法均能够较好地反映出混凝土试块内部的缺陷位置及分布。(3)针对基于射线理论的混凝土超声层析成像中由于射线稀疏、不完全造成的不适定问题,提出了一种基于菲涅耳体的混凝土超声层析成像方法。在混凝土超声层析成像中,超声波的能量传播实际上是在第一菲涅耳体内,利用菲涅耳体代替射线来进行混凝土层析成像,可以增加算法对混凝土测区内的覆盖范围,从而提高层析成像反演的速度和效率,并且层析成像结果图的分辨率和可靠性也将会得到提高。通过数值理论模型及混凝土试块试验验证了基于菲涅耳体混凝土超声层析成像方法的优越性。(4)研究了混凝土超声层析成像结果图像的后处理方法以及主要影响因素。对混凝土超声层析成像的结果图像采用谱系聚类分析进行处理。混凝土试块试验表明,谱系聚类分析图像后处理方法能够很好地抑制伪像,凸显缺陷,提高结果图像的分辨率、可靠性和可读性,从而增强结果图像的缺陷识别能力。对影响层析成像结果的多种因素进行试验研究分析,从而便于选取更加稳定、适合混凝土层析成像的方法。更多还原

【Abstract】 Because of the complexity of concrete material and deficiency in design, construction and maintenance, defects or damage of concrete components may emerge in construction or operation, which seriously affects the safety and even causes serious accidents. Therefore, it becomes particularly important to fast and accurately retrieve the inner quality condition of concrete components to provide a precise basis for evaluating concrete performance. In recent years, the ultrasonic computerized tomography technology was introduced in concrete nondestructive detection to display the internal structure, defect size and distribution of the concrete under test in the form of image accurately, vividly, and completely, which makes up limitations of traditional concrete detection method and becomes a new development trend of concrete structure detection. In this doctoral dissertation, my in-depth and systematic research in the aforementioned direction is introduced, including the principle of concrete ultrasonic testing, concrete ultrasonic tomography imaging theory and forward inversion method based on ray theory, concrete ultrasonic tomography method, image post-processing method for chromatography imaging results, and main factors influencing tomography results based on Fresnel volume, etc. The main research results are as follows:(1) An improved ray tracing method, based on Snell’s law and combining with the perturbation method, was proposed aiming at certain characteristics rules of random distributions of wave velocity and, defect within concrete, which was suitable for the forward characteristic of concrete ultrasonic computerized tomography. Test results of different models show that, the improved ray tracing path is more close to the real acoustic path. And the improved perturbation method for ray tracing based on Snell’s law can be applied in forward process of concrete ultrasonic computerized tomography imaging.(2) Through the study of iterative inversion algorithm for conventional concrete ultrasonic computerized tomography and the least square algorithm, two improved ART algorithms as well as the improved conjugate gradient algorithm and improved LSQR algorithm for concrete ultrasonic computerized tomography were proposed. Concepts of velocity classification and progressive block in imaging unit were introduced in the two improved ART algorithms. A preprocessing matrix was introduced in the improved conjugate gradient algorithm for projection matrix preprocessing to reduce the condition number of the projection matrix. A damping factor was added to the improved LSQR algorithm to suppress the impact caused by errors. The improved algorithms and the original algorithms were compared and analyzed via the numerical model and concrete block test. The results show that the proposed algorithm can improve the speed and accuracy of calculation. The algorithm effectiveness was verified by the imaging results analysis of two groups of concrete specimens’ internal defect. The imaging results show that the improved algorithm can better reflect the location and distribution of the defects in concrete specimens.(3) Aiming at the ill-posed problem caused by the Ray sparse and incompleteness in the concrete ultrasonic computerized tomography based on ray theory, a concrete ultrasonic computerized tomography method based on Fresnel volume was proposed. Ultrasonic energy propagation is actually inside the first Fresnel volume, so using Fresnel volume instead of ray to conduct Concrete Tomography would increase the coverage area of concrete test region, and thereby increase the speed and efficiency of the tomographic inversion, and improve the resolution and reliability of the tomography result figure. Numerical models and concrete specimen test proved the superiority of concrete ultrasonic computerized tomography method based on Fresnel volume.(4) The image post-processing of the concrete ultrasonic computerized tomography and the main influencing factors of ultrasonic computerized tomography were studied. Hierarchical clustering analysis was used in the image processing of concrete ultrasonic computerized tomography. Concrete specimen tests show that hierarchical clustering can suppress artifacts, highlight the defects and improve resolution, reliability, and readability of the resulting images. So it can improve the ability to identify defects through resulting images. In addition, a variety of influencing factors of ultrasonic computerized tomography was investigated and analyzed, which would facilitate more stable and suitable design of concrete ultrasonic computerized tomography.更多还原