顾及流变参数的时序InSAR公路形变模型研究

The precision of deformation estimation using time series InSAR technique is restrained by deformation model. Most of the traditional deformation models for InSAR deformation monitoring are pure empirical mathematical models, without considering the physical mechanism of the monitored object. Due to this, a new InSAR time series deformation model for soft clay road, considering physical parameters, is established, in order to improve the precision of long time deformation monitoring after road embankment settlement construction, and more reasonably interpret the displacement result. Firstly, with analysis of the displacement mechanism of the soft clay, the rheological models are analysed and compared, and the relational functional model between the LOS deformation and rheological parameters will be established; Secondly, the nonlinear characteristics will be studied, in the purpose of solving the complicated nonlinear parameter estimation problem, meanwhile the significance of the parameters in the new model will be tested meticulously, in order to optimize the rheological parameters in the new model; The influence relationship between the rheological parameters and the deformation estimations will be analysed, to show the error transformation discipline between the observation and parameters. Finally, case study of deformation estimation will be carried out, to assess the precision of modelling comprehensively. In this study, the rheological characteristics for soft clay road are considered in traditional InSAR deformation model, and the time series InSAR physical mechanical deformation model can be established; the result can be utilized to interpret the long-time deformation, which can be referred to by the field of road embankment stability control.

形变模型制约着时序InSAR技术形变估计的精度，传统InSAR形变模型大多采用一种或几种纯经验函数组合来拟合形变随时间演化的物理过程，未顾及监测对象物理力学特性，本项目拟针对软土地基公路变形监测，结合土体流变学，提出一种顾及物理参数的形变模型，以提高公路形变监测精度，更合理的解译形变估计结果。首先剖析软土地基公路变形机理，比选流变模型，建模InSAR视线向形变与流变参数的函数关系；其次，剖析新模型非线性特性，研究可顾及非线性强度的参数估计算法，实现新模型复杂非线性参数估计，并开展模型参数检验与优化；定量、系统分析各流变参数与形变、相位的影响关系，剖析其误差传递特性，以揭示流变参数对形变估值的影响规律；最后，开展公路测试区形变估计实例研究，验证模型精度。项目将结合软土地基公路材料流变特性，实现时序InSAR物理力学模型的构建；有利于公路工后长期变形分析解译，可为路基稳定性控制提供借鉴。

形变模型制约着时序InSAR技术形变估计的精度，传统InSAR形变模型大多采用一种或几种纯经验函数组合来拟合形变随时间演化的物理过程，未顾及监测对象物理力学特性，本项目针对软土地基公路变形监测，结合土体流变学，提出了顾及物理参数的形变模型。首先剖析软土地基公路变形机理，比选流变模型，建模InSAR视线向形变与流变参数的函数关系，构建了三种改进的时序InSAR流变模型；其次，剖析新模型非线性特性，提出一种基于正则化高斯牛顿迭代法与遗传算法串行的流变参数估计算法（GARN算法），实现了新模型复杂非线性参数估计；定量、系统分析了各流变参数与形变、相位的影响关系，剖析了新模型的误差传递特性。本项目对两个测试区分别开展了模拟实验和真实实验，获取了其流变参数解和形变序列，并辅助解译了测区形变成因及其变形规律。项目研究发现：新模型能够探测出传统形变模型无法探测出的由软土流变性而导致的非线性沉降；新模型获取的形变值与流变参数值呈反比的关系，即弹性模量和粘度系数的估计值越大，获取的形变量级越小；反之则形变量级越大，表明InSAR获取的流变参数值可以反应测区形变的时空分布特征；利用GARN算法精化模型参数，可弥补传统的非线性参数估计算法中迭代矩阵都需要进行求逆运算的缺陷，适用于迭代矩阵是病态矩阵或者奇异矩阵的情况，相比传统遗传算法在求解精度和效率上均有显著提升；通过与水准结果的比对分析发现，与传统线性模型相比，新模型结果与水准测量结果更加接近，精度有显著提升。本项目在传统InSAR模型中融入物理参数，深化了模型的理论意义；获取的流变参数估值，可为软土地基工程变形稳定性控制提供借鉴；研究成果提高了公路变形监测精度，辅助解译了软土地基公路工后长期变形分析，对预防地基失稳引发的安全事故具有重要的工程价值，可在我国软土地基区域进行推广应用。