湿陷性黄土地区高速铁路路堤下长短桩复合地基沉降理论研究

发布时间：2018-04-23 23:24

本文选题：湿陷性黄土 + 长短桩复合地基　；参考：《长安大学》2014年硕士论文

【摘要】：高速铁路的正常运营对路基沉降要求相当严格，其中最关键的是控制地基沉降然而，长短桩复合地基技术在湿陷性黄土地区应用后，不但控制了地基总沉降量，而且也取得了良好的经济效益和社会效益但是，前人对其理论研究还不够深入和完善因此，本文依托湿陷性黄土地区在建的高速铁路项目对长短桩复合地基的沉降进行系统的研究本文研究内容主要体现在以下几个方面：第一，分析长短桩复合地基在路堤荷载作用下的工作机理后，总结了长短桩复合地基承载力计算公式同时，概述了多种复合地基沉降计算方法，，并对其进行了分析第二，选取两个典型的路基断面，采用弹性土堤法Boussinesq三角形荷载法和Boussinesq组合荷载法计算得到地基附加应力，然后通过复合模量法和弦线模量法分别计算长短桩复合地基的沉降值，并与实测沉降值比较，得到了适用于湿陷性黄土地区长短桩复合地基的沉降计算方法第三，利用FLAC3D软件对长短桩复合地基进行数值模拟分析，桩间土采用Mohr-Coulomb模型，在数值计算过程中，最佳的深度取3.5b，最佳的宽度取4.5b，b为路基底宽度数值模拟结果表明，随着垫层厚度垫层模量的增大，地基沉降量分别表现出增加和减小；随着长桩长度和短桩桩端以上桩间土模量的增大，地基沉降量显著减小；随着长桩模量和短桩长度模量的增大，地基沉降量减小不明显第四，根据两个试验段路基沉降观测资料，运用双曲线法抛物线法和BP神经网络进行沉降预测其中，BP神经网络模型的构建，样本的训练以及预测借助MATLAB软件来完成然后，计算并对比三种预测方法的拟合指标r2和相对误差，发现采用BP神经网络最适合预测湿陷性黄土地区长短桩复合地基的沉降，其拟合精度和预测精度都很高
[Abstract]:The normal operation of high-speed railway requires the subgrade settlement very strictly, among which the key is to control the foundation settlement. However, after the application of the long-long pile composite foundation technology in the collapsible loess area, not only the total settlement of the foundation is controlled, And it has also achieved good economic and social benefits. However, the previous research on its theory is not deep enough and perfect. Based on the high speed railway project under construction in collapsible loess area, this paper studies the settlement of long-long pile composite foundation systematically. The research contents of this paper are mainly reflected in the following aspects: First, after analyzing the working mechanism of long-long pile composite foundation under embankment load, the calculation formula of bearing capacity of long-long pile composite foundation is summarized. At the same time, several calculation methods of composite foundation settlement are summarized and analyzed. Secondly, two typical subgrade sections are selected and the additional stress of the foundation is calculated by the elastic earth embankment method Boussinesq triangle load method and Boussinesq combined load method. Then the settlement value of long-long pile composite foundation is calculated by the composite modulus method and the string modulus method, and compared with the measured settlement value, the settlement calculation method suitable for the long-long pile composite foundation in collapsible loess area is obtained. Thirdly, the numerical simulation of long-long pile composite foundation is carried out by using FLAC3D software. The Mohr-Coulomb model is used in the soil between piles. In the process of numerical calculation, the optimum depth is 3.5b, and the optimum width is 4.5b. With the increase of cushion modulus, the settlement of foundation increases and decreases respectively, and with the increase of the length of long pile and the modulus of soil above the end of short pile, the settlement of foundation decreases significantly. With the increase of the modulus of long pile and the modulus of length of short pile, the settlement of foundation is not obvious. Fourthly, according to the observation data of subgrade settlement in two test sections, the hyperbolic parabola method and BP neural network are used to predict the settlement. The model of BP neural network is constructed, the training and prediction of the samples are completed by MATLAB software. The fitting index R2 and relative error of the three prediction methods are calculated and compared. It is found that BP neural network is the most suitable for predicting the settlement of long-long pile composite foundation in collapsible loess area, and its fitting accuracy and prediction accuracy are very high.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU470

【参考文献】