泛研网

公众版 查看说明
全球科研项目数据库2026版

基于跨孔雷达数据概率反演的地下连续墙缺陷识别方法研究

项目来源

国家自然科学基金(NSFC)

项目主持人

覃晖

项目受资助机构

大连理工大学

项目编号

41904095

立项年度

2019

立项时间

未公开

项目级别

国家级

研究期限

未知 / 未知

受资助金额

25.00万元

学科

地球科学-地球物理学和空间物理学-油气地球物理学

学科代码

D-D04-D0408

基金类别

青年科学基金项目

关键词

无损检测 ; 跨孔雷达 ; 地下连续墙 ; 探地雷达 ; 概率反演 ;

参与者AI

张东昊;覃晖;唐玉;谢雄耀;王峥峥

参与机构AI

大连理工大学;同济大学

项目标书摘要：地下连续墙是保障深大基坑施工安全的关键载体，但由于缺乏有效手段在基坑开挖前对地下连续墙进行检测，导致因地下连续墙缺陷造成的基坑安全事故时有发生。为此，本项目聚焦跨孔雷达检测方法，针对电磁波在复杂地下介质中的传播机理以及跨孔雷达数据反演等关键科学问题展开研究。通过测定地下连续墙、土体及各类缺陷的介电常数和电导率，获得地下介质的介电特性参数；利用时域有限差分方法进行三维数值仿真，分析不同频率电磁波的传播特性和复杂结构的影响方式，揭示电磁波在复杂地下介质中的传播规律；结合模型试验，挖掘各类缺陷的跨孔雷达数据特征；引入贝叶斯后验概率公式，结合马尔科夫链蒙特卡罗算法，考虑参数的先验信息和各类误差的影响，得出电性参数的后验概率分布。从而提高反演精度并定量给出反演结果的不确定性大小，实现对地下连续墙缺陷的准确识别。本项目的研究将为跨孔雷达方法在地下连续墙检测领域的成功应用提供理论基础和技术支撑。

Application Abstract: Diaphragm wall is a key factor to ensure construction safety for deep excavations.Yet for the lack of effective detection method,accidents often occur during construction due to diaphragm wall defects.Therefore,this study focuses on the crosshole ground penetrating radar(GPR)method for diaphragm wall defect detection,aiming at revealing the propagation mechanism of electromagnetic(EM)waves in complicated underground media,and improving the inversion method of crosshole GPR data.By measuring the permittivity and conductivity of diaphragm wall,surrounding soil and defects,dielectric properties of underground media can be obtained.According to 3D numerical modelling using the finite-difference time-domain(FDTD)method,the influence of different frequencies and complex structure on the EM wave propagation is analyzed and the propagation characteristic can be summarized.Combined with physical model experiment,crosshole GPR data features for different types of defects can be extracted.Based on the Bayesian Formula and Markov chain Monte Carlo(MCMC)algorithm,the posterior distribution of parameters can be obtained,taking into consideration prior information of parameters and errors of different sources.The result improves inversion accuracy and quantifies uncertainties,thus identifies diaphragm wall defects exactly.Therefore,this study provides theoretical and technical supports for successful application of crosshole GPR in diaphragm wall defect detection.

项目受资助省

辽宁省

项目结题报告(全文)

地下连续墙是保障深大基坑施工安全的关键载体，但由于缺乏有效手段在基坑开挖前对地下连续墙进行检测，导致因地下连续墙缺陷造成的基坑安全事故时有发生。本项目基于跨孔雷达方法，从电磁波在地下连续墙复杂介质中的传播规律出发，挖掘各类缺陷的跨孔雷达数据特征；研究跨孔雷达数据贝叶斯概率反演理论，提高对地下连续墙介电参数的反演精度，实现对地下连续墙缺陷的准确识别。具体研究内容及相关成果包括：(1)分析了跨孔雷达电磁波在复杂地下介质中的传播特性，得出跨孔雷达方法可用于地下连续墙病害检测的条件。一是地下连续墙结构为低损耗介质件，电磁波能在地下连续墙结构内传播；二是病害部位由于含水量增大导致介电常数显著增加，使得病害部位和完整结构出现明显的电性参数差异，这种差异使电磁波发生反射、折射和散射等现象，因此可通过对电磁波的分析获得病害信息。(2)通过数值模拟总结了地下连续墙缺陷的跨孔雷达数据特征。层状病害可通过其在零高差数据中的特征，即直达波走时不改变但振幅显著减小，来判断层状病害的发生范围。块状病害的零高差初至走时在病害处发生延迟、直达波振幅明显减小，病害具体位置和分布可由多高差数据反演得出。接头楔形间隙零高差初至走时随深度增大逐渐增加，直达波振幅随深度增大逐渐减小。槽底淤泥零高差初至走时突然发生较大延迟，直达波振幅出现“断崖式”减小，淤泥的具体分布范围可由多高差数据反演得出。(3)建立了基于贝叶斯概率反演的地下连续墙缺陷识别方法。该方法基于贝叶斯定理，建立观测数据和模型参数间的概率联系。采用马尔科夫链蒙特卡罗(MCMC)模拟结合DREAM(ZS)算法来实现参数的后验分布采样，该算法构造马尔科夫链，并通过差异进化算法和Metropolis-Hastings(MH)准则使马尔科夫链收敛于后验分布。与确定性反演方法相比，该方法能够显式处理测量误差，并且对反演结果的不确定性做出定量判断，反演结果更加客观和可信。

  • 排序方式：
  • 1
  • /

  • 1.Generative Adversarial Network Based Inversion of Cross-hole Radar Data

    • 关键词：
    • Deep learning;Finite difference time domain method;Radar imaging;Cross-hole radars;Data interpretation;Deep learning;Generative adversarial network;Interpretation methods;Inversion accuracy;Lower resolution;Network-based;Radar data;Subsurface structures
    • Zhang, Donghao;Tang, Yu;Qin, Hui;Wang, Yuanzheng;Yao, Yao;Zhang, Lin
    • 《10th International Conference on Environmental and Engineering Geophysics, ICEEG 2023》
    • 2023年
    • June 7, 2023 - June 12, 2023
    • Beijing, China
    • 会议

    Cross-hole radar is a popular method to characterize subsurface structures, yet traditional data interpretation methods have limitations: tomography method has poor inversion accuracy, while the full waveform inversion method costs huge computing time. In order to solve the above problems, an automatic inversion algorithm based on generative adversarial networks (GAN) is developed in this paper to interpret the permittivity from cross-hole radar B-scan images. This algorithm uses a low-resolution GAN to extract the global features in the cross-hole radar data to reconstruct the dielectric constant distribution map with low resolution, and then adopts a high-resolution GAN to enhance the resolution of the inversion results. The algorithm is trained on 1000 pairs of cross-hole radar data obtained from the finite-difference time-domain (FDTD) method. Finally, 100 pairs of similar data which has never been shown in the network are used to verify the inversion performance of the algorithm. The results show that the inversion accuracy of is greater than 90%, and the structural similarity index measure (SSIM) of the reconstructed image reaches 0.9. In addition, the proposed method also has rapid computing speed. © Published under licence by IOP Publishing Ltd.

    ...

  • 3.MCMC approach for shield tunnel grouting layer estimation using ground penetrating radar

    • 关键词：
    • Markov processes;Monte Carlo methods;Ground penetrating radar systems;Geological surveys;Concrete construction;Electric conductivity measurement;Shielding;Mortar;Permittivity;Bayesian inversion;Ground Penetrating Radar;Ground penetrating radar (GPR);Inversion results;Markov chain monte carlo simulation;Model parameters;Probabilistic inversion;Relative permittivity
    • Qin, Hui;Tang, Yu;Wang, Zhengzheng;Xie, Xiongyao;Zhang, Donghao
    • 《China Rock 2020 Conference》
    • 2020年
    • October 23, 2020 - October 26, 2020
    • Beijing, China
    • 会议

    Ground penetrating radar (GPR) has been suggested as an effective tool for evaluating the grouting layer behind shield tunnel linings, yet the estimation of the grouting layer thickness is usually difficult. In this paper, we propose a probabilistic inversion method to evaluate the grouting layer using GPR images. This method uses a sliding window along the GPR scan axis combined with a Markov chain Monte Carlo (MCMC) simulation with Bayesian inversion to infer the grouting layer thickness together with the relative permittivity and electric conductivity. We illustrate this approach using a synthetic GPR experiment that simulates grouting layer detection in a shield tunnel along the longitude direction. A sliding window with a width of 0.2 m is used to estimate the model parameters and is moved along the scan axis with a step size of 0.2 m after each inversion. The results demonstrate successful estimation of the grouting layer thickness and its relative permittivity and electric conductivity by the proposed method. Moreover, this method is capable of quantifying uncertainties in the inversion results.
    © Published under licence by IOP Publishing Ltd.

    ...
  • 排序方式：
  • 1
  • /