节点文献
土木工程的层析成像与广义反演研究
Study on the Computer Tomography and Generalized Inversion of Civil Engineering
【作者】 王振宇;
【导师】 刘国华;
【作者基本信息】 浙江大学 , 防灾减灾与防护工程, 2003, 博士
【摘要】 在综合分析土木工程层析成像研究历史和现状的基础上,系统深入地研究了土木工程层析成像与广义反演的新理论、新方法和若干实用技术,研制了层析成像软件,数值仿真、室内试验和工程实例证明了本文方法的科学性、可行性和应用效果。研究成果可为结构的健康诊断与安全评估提供一定的科学依据。 本文的研究结论和创新点如下:(1)提出基于广义反演的层析成像研究框架,该框架包括外业测试与评估、前处理、正反演求解和结果分析评价等内容。(2)对层析成像所面临的离散不适定问题进行研究,通过对奇异值谱和Picard图的分析,剖析离散不适定问题求解困难的本质。(3)对常规SIRT算法进行改进;正则化方法在残差范数和解的范数之间进行最优折衷的思想对其他广义反演方法有较好的借鉴意义。(4)提出基于广义逆的一类反演方法和解评价指标。在分析解估计评价指标与阻尼系数关系的基础上,提出阻尼类反演方法的阻尼系数优选技术,采用多目标优化的评价函数法进行求解,该阻尼系数优选技术对初始值不敏感,所得阻尼系数合理。(5)研究三种误差对反演结果的影响和小波分析的降噪技术;引入向后扰动的概念,分析广义逆算法的稳定性;提出对先验信息和约束条件的处理方法;在同等测试条件下,广义逆方法具有更高的识别精度。(6)提出可用于层析成像异常区识别的若干个定量算法,开展混凝土构件层析程成像试验,研究表明层析成像的缺陷评估更直观、准确、信息量更丰富,比其他常规方法有独特的优势。(7)电阻率层析成像正演算法采用点源二维电场的有限单元法,反演算法采用平滑度约束的最小二乘法,以均方根误差作为目标函数。(8)采用改进的Miller Soil Box进行土样的电阻率试验,根据四因素三水平正交试验得出影响土的电阻率变化的主次因素顺序是:含水率、孔隙水的导电性、饱和度、土的种类。提出一个基于推广阿尔奇公式的粘土电阻率模型。对实测结果进行电阻率层析成像,能较好的发现地下目标,结合粘土的电阻率模型将有助于检测土层的含水率。
【Abstract】 Several new theory, key technica and practical methods of computer tomography and generalized inversion on civil engineering have been studied comprehensively and systemically to solve the current bootleneck problems existed in computer tomography on the basis of a synthetically review of its research history and state-of-the-art. A software of computer tomography system is developed. Moreover, Numerical simulations, indoor experimentations and practical examples are adopted to elaborate the scientificity. feasibility and application effectiveness of the research result, which also provide some scientific foundations of safety assessment, health monitoring and damage diagnosis for structres in civil engineering.The main contents and the conclusions of the study are described as follows:(1) The framework and methodology of computer tomography basised on generalized inversion is put forward. The framework include outdoor test and assessment, pre-processing, forward modelling and generalized inversion, result analysis and evaluation.(2) Discrete ill-posed problems confronted by computer tomography is studied. By the means of singular spectrum and picard plot, the essence of the difficulty to solve discrete ill-posed problems is analyzed.(3) The SIRT method is improved in this paper. The heart of regularization method is the optimum compromise between the solution norm and the corresponding residual norm, which is very meaningful to the other generalized inversion methods.(4) A series of generalized inversion methods and solution assessment index based on generalized inverse of matrix are proposed. The multi-object optimum method of evaluation function is adopted to sloved such an optimum problem and confirm the damp coefficient. The optimum methodology is not sensitive to initial value, and can always achieve a sensible solution.(5) The perturbation of three kinds of noise on the inversion solution is studied and wavelet analysis is introduced to reduce the noise. Backward error and backwardperturbation analysis is also discussed to analyze the computational stability of generalized inversion. Moreover, a practical method to take into account the transcendental information is put forward. And. it is also showed that generalized inversion methods have a better identifying precision under the same situation.(6) Several quantitive methods are propsed for identifying the velocity abnormity in tomography. The indoor test shows these flaw evaluation methods are more intuitionistic and exact than the other NDT methods.(7) The two dimension electrical resistivity tomography (ERT) uses the finite element method to carry out the forward model. The inversion routine used by the program is based on the smoothness-constrained least-squares method. Root mean squared error is the object function in the optimum problem .(8) Miller soil box is ameliorated and adopted to perform a standardized laboratory procedure for measuring soil electrical resistivity. Through orthogonal test, it is discovered that the influences of electrical resistivity are decreased in such an order: water content, pore-water electrical conductivity, degree of saturation, kind of soil. Then, an electrical resistivity model and its Archie’s parameters of typical clay are established. Finally, A practical demonstration is given to illustrate the ability to find potential object by ERT. Combined with the electrical resistivity model. ERT will redound to forecast the water content of the soil.