【作者】 蒋超；

【导师】 夏元友；

【作者基本信息】 武汉理工大学 ， 岩土工程， 2004， 硕士

【摘要】 云南祥临公路沿线工程地质条件复杂，岩体风化严重，岩土性能差，易滑地层分布广，已发生滑动的高边坡有52处之多，并且由勘查发现有大量存在滑动变形危险的古滑坡。本文将数据库技术、模糊数学、神经网络、计算机技术等学科与边坡工程研究交叉综合，建立云南祥临公路高边坡及滑坡数据库系统；采用随机和随机模糊法对祥临公路沿线滑坡的滑带土(主要岩性为含碎石粉质粘土)的抗剪强度指标进行统计：分析，给出其取值范围；最后，建立具有此类软弱面且最危险滑面为三段直线的岩质边坡模型，利用神经网络及正交设计，分析此类边坡的最优稳定开挖坡角的取值范围，建立计算最优稳定开挖坡角的人工神经网络模型以提供一种快速计算最优稳定开挖坡角的途径。论文主要工作如下： 1)根据祥临公路实际情况和资料的详尽程度对祥临公路沿线边坡进行分类，借助VC#．net及SQL Server建立了高边坡及滑坡数据库系统，该系统能让公路管理人员及边坡监测研究人员通过网络方便直观的获得并修改祥临公路沿线边坡的具体信息； 2)对祥临公路滑坡中岩性为含碎石粉质粘土的滑带土(分为滑动前以及滑动后两种状态)抗剪强度指标分别用随机和随机模糊两种统计方法进行分析，得到其抗剪强度指标(c，φ′)值的两组概率密度函数及取值范围，确定在滑动前以及滑动后粘聚力c及内摩擦角φ的建议值分别为[18.563 kPa，22.856kPa]、[18.172°，23.779°]以及[18.323 kPa，22.475 kPa]、[8.824°，11.453°]；最后采用传统法、正交变换法、经验概率拟合法及随机模糊法等4种不同方法对某个具体滑坡滑带土抗剪强度指标的试验值进行统计分析； 3)在假定滑动面为三段直线的情况下，建立了含弱面岩质边坡的稳定坡角优化模型并确定了影响最优坡角的7个变量因素(即软弱面上的内摩擦角φ、粘聚力c、重度γ、稳定系数F、自然坡角α、软弱面位置控制参数σ及L)；基于数据库统计结果，采用正交设计确定祥临公路此类边坡最优稳定开挖坡角的建议范围为28～52°，以及影响最优稳定开挖坡角的因素敏感性关系(软弱面位置控制参数的影响最大，其次是内摩擦角，再次是粘聚力，最后才是自然坡角)；通过对正交设计成果的学习，建立BP神经网络(ANN)模型，为具体滑坡最优稳定开挖坡角的确定提供一个简便的解决途径；最后对祥临公路某个具体滑坡做出分析以证明运用上述ANN模型进行边坡稳定开挖坡角优化分析的可行性。更多还原

【Abstract】 The geological conditions in Xiang-Lin Road in Yunnan is complex and the performance of the rock and soil is poor, so the slopes, which contain the structural plain and whose gliding mass always are clay containing the reduced stone, distribute largely alone the road. The number of the high slopes having slide is up to 52, and there are many ancient slopes, which have been found out to have slide probability though the survey. In this paper, the several disciplines (database, neural networks, fuzzing mathematics, computer science, etc.) and the slope engineering have been synthesized, and the database of the high slope in Xiang-lin Road is established .In the basis of the database, the test data of the shear strength of the gliding mass, whose component are the clay containing the reduced stone, are analyzed in random statistic method and random-fuzzy method, and the spans of the shear strength of the this gliding mass are given to be indicator for the future research. At last, a model, which represents the rock slope containing the weak surface, is created. Based this model, using the orthogonal design and the neural networks, the span of the optimum angle of the slopes and sensitivity of the variable is found to guide the practice. Several aspects have been studied in this paper, including:1) According to the practical situation of the Xiang-Lin Road and the details of the survey report, using the VC# and SQL Server, the database is established which is helpful to inquire and modify the data for the managerial staff and the researchers monitoring the slope through the Internet.2) Based the database of the slope, the test data of the shear strength of the gliding mass, are analyzed in random statistic method and random-fuzzy method. The density function and the span of the shear strength of the previous gliding mass are worked out, and the shear strength of the specific slope is gained in four ways.3) The model of this kind of the rock slopes is established on the assumption that the potential slide surface of the slope is consisted of three lines, and seven parameters influencing the choices of the angle of the slope are determined. Using the orthogonal design, the span of the optimum angle of the previous slopes and the sensibility of the five variables is worked out. By means of the neural networks, a nonlinear relation between the seven parameters and the optimum angle of the slope is established, in the basis of which the optimum angle of the specified slope is worked out to prove the correctness of the relation.更多还原