基于神经网络的高边坡稳定性预测与加固研究

发布时间：2018-04-24 06:57

本文选题：边坡 + 破坏机理　；参考：《中国矿业大学》2015年硕士论文

【摘要】：边坡稳定性问题作为岩土工程研究领域中的重要课题,广泛涉及到公路、铁路、水利工程、建筑基坑、煤矿等基础建设中。边坡稳定性问题主要研究:边坡稳定性的预测与加固效果预测。通过对边坡稳定性的预测与加固预测,可以快速、直观地判断出边坡的状态。边坡加固效果模拟可以近似仿真现场加固效果,低成本获得较优的边坡加固方案,从而为边坡加固方案的现场实施提供一定的科学参考。本文基于理论分析和数值模拟,系统分析了考虑渗流场作用下高边坡的稳定性与加固效果,主要完成了以下几个方面的研究工作:(1)系统概述了岩土屈服准则和岩土微观破坏机理、边坡破坏的判别准则以及边坡强度折减法等基本理论。(2)研究了边坡在自重作用下的变形、塑性区、安全系数等特征,并且进一步研究了边坡在渗流场、应力场耦合作用情况下的稳定性和破坏规律。为了简化预测系统,详细对比了渗流场作用下和代替重度法的边坡稳定性差异,为建立有效的边坡稳定性预测系统提供了理论基础。(3)基于神经网络理论,对比了几个训练函数的准确性,并且确定了几个影响因素(重度、内聚力、内摩擦角、坡角、坡高等)作为输入层,建立了5*12*1的神经网络预测高边坡稳定性的诊断平台。通过大量的边坡数值模拟结果作为训练样本,并且验证了神经网络的准确性;并在以上基础上,对诊断平台进行深入研究,建立了边坡加固效果神经网络预测系统,确定了多个影响因素(重度、内聚力、内摩擦角、坡角、坡高、加固前安全系数、土钉长度、土钉间隔、土钉角度、网喷厚度等)作为输入层,建立10*16*1的神经网络诊断平台,将大量规范中的加固方案作为训练样本,并且验证其准确性。(4)结合上述神经网络诊断平台,先对徐济高速公路边坡进行稳定性预测,然后通过边坡加固神经网络预测制定加固方案,最后通过Flac3D模拟加固方案,详细了解边坡受力状态并对预测结果验证。
[Abstract]:As an important subject in the field of geotechnical engineering, slope stability is widely involved in the infrastructure construction of highway, railway, water conservancy engineering, building foundation pit, coal mine and so on. Slope stability is mainly studied: slope stability prediction and reinforcement effect prediction. Through the prediction of slope stability and reinforcement prediction, the state of slope can be judged quickly and intuitively. The simulation of slope reinforcement effect can approximate the field reinforcement effect, and obtain the better slope reinforcement scheme at low cost, thus providing a certain scientific reference for the field implementation of the slope reinforcement scheme. Based on theoretical analysis and numerical simulation, the stability and reinforcement effect of high slope considering seepage field are systematically analyzed in this paper. The main work of this paper is as follows: (1) A systematic overview of the yield criterion of rock and soil and the mechanism of microcosmic failure of rock and soil, the criterion of slope failure and the basic theory of slope strength reduction. (2) the deformation of slope under the action of self-gravity is studied. The characteristics of plastic zone and safety factor, and the stability and failure law of slope under the coupling of seepage field and stress field are further studied. In order to simplify the prediction system, the difference of slope stability under seepage field and in place of severe method is compared in detail, which provides a theoretical basis for the establishment of an effective slope stability prediction system based on neural network theory. The accuracy of several training functions is compared, and several influencing factors (heavy, cohesion, angle of internal friction, angle of slope, height of slope, etc.) are determined as input layers, and a diagnostic platform for predicting the stability of high slope by neural network is established. Through a large number of slope numerical simulation results as training samples, and verify the accuracy of neural networks, and on the basis of the above, the diagnosis platform is studied, and the slope reinforcement effect neural network prediction system is established. Several factors (heavy, cohesive, internal friction angle, slope angle, slope height, safety factor before reinforcement, soil nail length, soil nailing interval, soil nail angle, net spray thickness, etc.) were determined as the input layer to establish a neural network diagnostic platform of 10 / 16 / 1. Taking a large number of reinforcement schemes in the code as training samples and verifying its accuracy, combined with the above neural network diagnostic platform, the slope stability of Xuji Expressway is forecasted. Then the reinforcement scheme is established through the prediction of the slope reinforcement neural network. Finally, the stress state of the slope is understood in detail and the prediction results are verified by the Flac3D simulation reinforcement scheme.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TU43;TP183

【参考文献】