基于BP神经网络的隧道围岩参数反分析

发布时间：2018-06-25 15:11

本文选题：隧道监测 + 围岩位移　；参考：《西安工业大学》2014年硕士论文

【摘要】：长期以来,对隧道工程进行准确的设计,施工一直是隧道建设的一个最为重要的步骤,围岩的性质状态对于施工和设计起着决定性的作用,几十年来,岩土工程师们一直在研究如何能够准确合理地获得表征隧道围岩力学性状的参数。尤其是近十年来,新奥法的施工工艺得到极大的推广,监测信息的反馈技术也得到相应发展,这样的工艺使得通过反分析方法来确定围岩力学参数的研究更加有意义。本文以十天高速公路茶镇隧道施工监测为研究对象,以现场变形量测数据为基础,通过一系列数学方法进行回归模拟,用BP神经网络构建了力学参数反分析的模型。以茶镇隧道现场实际量测拱顶沉降及周边收敛的数据为研究对象,以专业数学软件MATLAB7.0内置神经网络工具箱作为手段,利用构建起来的模型和正算模拟软件两者互相结合,继而选用正交设计的方法选取模型所需要的学习训练样本。在对这些样本进行了充分的学习和训练之后,从而可以建立隧道围岩参数与隧道位移(水平收敛位移和拱顶沉降位移)之间的对应的非线性映射关系。继而将所建立的这种呈非线性状态的映射关系应用到对茶镇隧道的三个参数弹性模量E和粘聚力c以及内摩擦角φ进行反演分析,最终得出合理取值。本文研究的基于BP神经网络的位移反分析方法对隧道工程设计、施工、监测提供一定的决策依据,对隧道后续开挖的设计、施工、监测提供建议值,实现信息的反馈,对相似隧道工程提供指导和借鉴意义。
[Abstract]:For a long time, the accurate design and construction of tunnel engineering has been the most important step in tunnel construction. The nature of surrounding rock plays a decisive role in construction and design. Geotechnical engineers have been studying how to accurately and reasonably obtain the mechanical properties of tunnel surrounding rock. Especially in the past ten years, the construction technology of the New Austrian method has been greatly popularized, and the feedback technology of monitoring information has been developed accordingly, which makes the study of determining the mechanical parameters of surrounding rock by the method of back analysis more meaningful. In this paper, the construction monitoring of Cha Zhen Tunnel of Shitian Expressway is taken as the research object, based on the field deformation measurement data, regression simulation is carried out by a series of mathematical methods, and the back analysis model of mechanical parameters is constructed by using BP neural network. Taking the data of actual measurement of dome settlement and its peripheral convergence in the field of Cha Zhen Tunnel as the research object, taking MATLAB 7.0 as a means of built-in neural network toolbox, this paper combines the constructed model with the positive simulation software. Then the orthogonal design method is used to select the learning training samples needed by the model. The nonlinear mapping relationship between tunnel surrounding rock parameters and tunnel displacement (horizontal convergent displacement and arch settlement displacement) can be established after the study and training of these samples. Then, the nonlinear mapping relationship is applied to the inversion analysis of the three parameters of elastic modulus E, cohesion C and internal friction angle 蠁 of Cha Zhen Tunnel, and a reasonable value is obtained. The back analysis method of displacement based on BP neural network in this paper provides a certain decision basis for tunnel engineering design, construction and monitoring, and provides the suggested value for the design, construction and monitoring of tunnel subsequent excavation, so as to realize the feedback of information. To provide guidance and reference for similar tunnel engineering.
【学位授予单位】：西安工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP183;U451.2

【参考文献】