【作者】 周亮；

【导师】 李刚；

【作者基本信息】 湖南大学 ， 土木工程， 2021， 硕士

【摘要】 边坡稳定评价是滑坡预防与治理的重要环节。边坡工程中,一方面由于边坡自身的复杂特性,另一方面,受分析模型与方法影响,其稳定表现出不确定性。随着科技的发展,人们对边坡稳定分析理论的精度与求解效率提出更高的要求,而一些传统方法存在局限。为此,本文以边坡稳定分析理论的计算精度与求解效率为目标,基于已有研究成果,从理论分析和数值模拟两方面,对边坡临界滑裂面搜索的优化算法和基于随机场的边坡可靠度分析等一些问题展开了研究,主要工作如下:首先,将混沌算法引入到搜索者寻优算法(SOA)中以增强算法的局部搜索能力,提出改进的搜索者寻优算法(ISOA),经过函数优化测试,ISOA的精度,效率与稳定性都有所提高。将ISOA同有限元滑面应力法结合,应用于边坡临界滑裂面搜索,编制出相应程序,通过两道ACADS考核题,验证了ISOA在边坡工程中适用性。其次,基于随机场理论,根据岩土参数的均值,方差,波动范围,相关函数等统计特征,建立岩土体参数对数随机场模型,采用KL级数展开法离散随机场,并分析一些参数对随机场离散精度的影响。最后,在随机场模型基础上,采用有限元滑面应力法计算边坡安全系数,以随机场离散后的独立标准随机变量为输入量,安全系数为响应量建立代理模型。考虑到样本点的维数较高,对样本点首先采取切片逆回归(SIR)降维,将降维后的样本点及其响应通过Kriging代理模型建立近似显示关系,再采用蒙特卡洛法,大量抽样代入代理模型求得边坡失效概率,并分析了波动范围、互相关系数和变异系数对边坡失效概率的影响。更多还原

【Abstract】 The stability assessment of slope is an important step of landslide forecast and treatment.In slope engineering,on the one hand,there are many complex characteristics in slope,on the other hand,slope is affected by analytical model and method,so its stability shows uncertainty.With the development of the science and technology,there are more requirments for the precision and solution efficiency of slope stability analysis theory,but some traditional methods have limitations.To this end,aiming at the accuracy and solution efficiency of slope stability analysis theory,on the basis of the existing research achievement,the optimization algorithm of critical slip surface and slope reliability analysis method based on the random field model are studied in this article by theoretical analysis and numerical simulation.The main task is presented as follows:Firstly,in oder to the enhance the local search ability,the improved seeker optimization algorithm(ISOA)which combined chaos algorithm is proposed.Through functions optimization,ISOA has a promotion on accuracy,stability and efficiency.Combined with finite surface stress method,improved seeker optimization algorithm is applied to seach the critical slip surface.And the corresponding procedure is coding.Through two test examples of ACADS,it can be verified that ISOA is suitable for application in slopes.Secondly,based on the random field theory,according to the mean variance,correlation distance,correlation function and other statistical characteristics of geotechnical parameters,the random field model is established.And the random field is discreted by Karhunen-Loveve(KL)expansion method.The influence of different parameters on the discretization precision of random field is analyzed.Finally,on the basis of the random field model,the safety factor of slope is calculated by finite surface stress method.The agent model is established between the input that is independent standard random variable in the random field and the response that is factor of safety.Given the high dimension of sample points,slice inverse regressive(SIR)method is used to reduce the dimension.Then Kriging agent model is established between the sample points whose dimension is reduced and its response.With a large number of new samples,the Monte-Carlo method is used to obtain failure probability of the slope.And the influence of correlation distance,related coefficient and coefficient of variation on the failure probability of slope is analyzed.更多还原