【作者】 罗辉；

【导师】 杨仕教；

【作者基本信息】 南华大学 ， 采矿工程， 2010， 硕士

【摘要】 我国是一个地震多发的国家,由地震引发的滑坡事故造成了人民生命财产的巨大损失。由于地震荷载的随机性和边坡体材料参数的不确定性,边坡体稳定可靠性在地震作用的全过程是动态变化的,因此,提出边坡体时程动力可靠性概念并研究了基于时程动力可靠性的边坡工程反演设计方法。本文主要研究内容和研究成果如下:1、同时考虑岩土体材料参数与地震荷载的随机性,分析边坡体在地震作用全过程下的可靠性变化规律,提出了动力时程可靠性概念。2、根据材料参数变异性采用正交试验设计方法对边坡体进行动力有限元计算,分析边坡体的峰值速度场分布规律及的动力响应变化特征。3、应用神经网络建立了边坡体随机参数的随机动力响应模型。以动力有限元和神经网络为基础,建立了有限元-神经网络-蒙特卡罗法(FEM-ANN-MCS)、有限元-神经网络-响应面法(FEM-ANN- RSM)、有限元-神经网络-最大熵密度法(FEM-ANN-MEA),有限元-神经网络-最佳平方逼近法(FEM-ANN-LSA)的四种时程动力可靠性的计算方法。通过某一土质边坡工程算例,分析了四种时程动力可靠性计算方法的优缺点和适用条件。4、建立了基于FEM-ANN-RSM时程动力可靠性的边坡工程锚杆长度反演设计方法,和以枚举法为思想的基于FEM-ANN-MCS时程动力可靠性的边坡工程锚杆反演设计方法。以某一锚杆支护边坡工程为算例,应用这两种方法进行了锚杆长度反演设计和分析。研究结果表明时程动力可靠性能较好的描述地震作用全过程边坡体可靠性变化情况,建立的FEM-ANN-MCS、FEM-ANN-RSM,FEM- ANN-MEA的边坡时程动力可靠性计算方法,计算精度较高,均可用来进行边坡体地震动力稳定性分析。其中,采用FEM-ANN-MCS法进行边坡支护锚杆长度反演设计更为适用。更多还原

【Abstract】 China is an earthquake country. Landslides triggered by an earthquake result in great losses of life and property. Because of the randomness of earthquake loading and uncertainty of slope material parameters, the reliability of the slope stability changes dynamically during an earthquake. Therefore, the concept of dynamic reliability for time-history is proposed and the inverse design method for slope works is developed based on this concept. The main contents and the research results of this thesis are listed as follows.1、The time-history dynamic reliability concept is proposed to analyze the changes in reliability of slopes during the entire process of earthquake action, in which the randomness of both geotechnical material parameters and earthquake loading is taken into account.2、The dynamic finite element analysis of the slope is performed using the orthogonal test design method according to the variability of the material parameters to analyze the peak velocity distribution and the characteristics of dynamic response.3、The random dynamic response model is established for Stochastic parameters of a slope using artificial neural network. Four methods of calculating time-history dynamic reliability are developed based on dynamic finite element and artificial neural network,which include finite element-artificial neural network-Monte Carlo method(FEM-ANN-MCS)、finite element-artificial neural network-response surface method (FEM-ANN-RSM)、finite element-artificial neural network-maximum entropy approximation ( FEM-ANN-MEA )、finite element-artificial neural network-least square approximation(FEM-ANN-LSA). The merits, shortcomings and applicability of the four methods are investigated through an engineering example of an actual soil slope.4、The inverse design method for anchors used in slope works is developed based on the proposed FEM-ANN-RSM method. The inverse design method is also developed based on the proposed FEM-ANN-MCS method using the logic of the enumeration method. These two methods are applied to an actual anchored slope to determine the length of anchors.Results of the investigation indicate the proposed time-history dynamic reliability can describe well changes in the reliability of slope during the entire seismic process. The proposed methods for calculating time-history dynamic reliability (FEM-ANN-MCS、FEM-ANN-RSM,FEM- ANN-MEA ) have high precision and can be used to evaluate seismic slope stability. The FEM-ANN-MCS method is more applicable in the inverse design of anchored slope works.更多还原