基于RS-PCA-GA-SVM的砂土液化预测方法研究

砂土液化是一种危害性比较大的自然灾害,对砂土液化进行判定预测在地质灾害防治领域中有重要的研究意义。通过粗糙集理论(Rough Set,RS)对影响砂土液化的6个初始评价指标(包括震级、土深、震中距、地下水位、标贯击数和地震持续时间)进行属性约简,去掉冗余或干扰信息,得到基于4个核心预测指标的数据集。通过主成分分析法(Principal Component Analysis,PCA)从核心评价指标中提取出主成分,采用支持向量机(Support Vector Machine,SVM)对数据集进行训练,用遗传算法(Genetic Algorithm,GA)优化参数,建立砂土液化的RS-PCA-GA-SVM预测模型。并结合砂土液化实际数据将预测结果与基于Levenberg-Marquardt算法改进的BP神经网络模型(LM-BP)的预测结果做比较。实例计算表明:基于RS-PCA-GA-SVM模型得到的砂土液化预测结果精度较LM-BP神经网络有很大的提高,判别结果与实际情况比较吻合,可在实际工程中应用。     

基于支持向量机的砂土液化预测分析

将支持向量机方法应用于砂土地震液化预测问题.考虑影响砂土液化的因素,选用震级、标贯击数、相对密实度、土层埋深、地震历时、地面运动峰值加速度和震中距7个影响因子作为液化判别指标,建立了砂土液化预测的支持向量机模型.以砂土液化实测数据作为学习样本进行训练,建立相应函数对待判样本进行分类.研究结果表明:支持向量机模型分类性能良好,是砂土地震液化预测的一种有效方法,可以在实际工程中进行推广.     

基于遗传神经网络的砂土液化判别模型

针对BP人工神经网络具有易陷入局部极小等缺陷，本文提出了将遗传算法与神经网络相结合，同时优化网络结构与权值、阈值的思想。根据地震液化的实测资料，建立了砂土液化判别的遗传神经网络模型，比较计算结果证明了该模型的科学性、高效性。文中并进行主成分分析，提出液化影响的主要因素。     

基于最小二乘支持向量机的砂土液化预测方法

使用最小二乘支持向量机分类方法建立了两个砂土液化预测模型,预测结果与野外实际情况全部相符,表明该分类方法用于预测砂土液化是可行的,且预测准确率高。     

关于砂土液化判别的若干意见

简要回顾了我国抗震规范中液化判别方法的发展历史,并对我国若干规范中的液化初步判别的条文进行了比较,提出了修改意见;对我国有代表性的几种液化判别方法及基于Seed简化方法修正的美国NCEER法进行了评述和综合对比,指出了我国现行若干规范中有关液化判别规定方面的错误;此外,还将GB50011－2001《建筑抗震设计规范》液化判别公式转变为类似于NCEER法的表达式,拓宽了该规范方法的应用范围。最后,结合某长江大桥工程的地基液化判别工作,探讨了深层土液化判别的可能性,结果表明,本文建议的方法是可行的。     

基于遗传神经网络的地震砂土液化判别研究

针对BP人工神经网络具有易陷入局部极小等缺陷,本文提出了将遗传算法与神经网络结合,同时优化网络结构的权值与阈值的思想,建立了砂土液化判别的遗传神经网络模型。根据地震液化的实测资料,分别对BP〗神经网络判别结果和遗传神经网络判别结果进行了比较,结果表明后者比前者判别能力要好些。     

饱和砂土地震液化判别的分形插值模型

借鉴分形基本理论,提出了基于分形插值模型的饱和砂土地震液化判别方法.该方法首先选取影响饱和砂土地震液化判别的7个主要因素,根据分类标准,采用在每级标准中随机内插的方法,得到40个标准样本,用于构建饱和砂土地震液化判别的分形插值模型;其次根据最大似然分类原则确定每个饱和砂土地震液化判别指标的评价分维数;然后利用加权求和法计算样本的综合评价值,并根据样本综合评价值与经验等级之间的关系建立分形插值评价模型;最后,进行了实例分析结果表明:该模型的评价结果合理、客观,计算得到的每个样本具体得分值,即使对属于同一级的样本也可以给出其地震液化程度的顺序,为饱和砂土地震液化评价工作提供了一种新的研究方法与思路.     

应用剪切波速判别砂土液化的研究综述

首先介绍了应用剪切波速判别饱和砂土振动液化的原理；然后从临界剪应变取值、砂土的抗液化能力与剪切波速相关关系的研究方法与设备、用剪切波速进行扰动评价、建立适用于不同土类的统一判别式四方面总结了国内外学者在该领域的一些研究成果及存在问题；最后针对以上问题提出作者的一些看法。     

唐山地震时不同烈度区液化砂土的特征及其液化判别

本文根据唐山地区的东、中、西三条剖面上的地震喷砂的物质成分和理化性质等分析,探讨了不同烈度区液化砂土的颗粒成分、化学成分、物理性质及砂颗粒形态等变化规律,并根据这些规律,编制出以物质成分和性质为基础的、具有多种因素的砂土液化判别图。该图可供砂土液化预测、砂土液化研究者参考     

基于Finn本构模型的饱和砂土地震液化分析

天然地震作用下的饱和砂土液化问题是岩土地震工程研究的重要课题之一。目前,国内推荐使用的规范法是基于实际地震液化调查而建立的判别方法,方法本身缺乏理论基础。采用Finn液化本构关系建立了砂土液化数值分析模型,运用有限差分法的动力时程分析模块,分析了饱和砂土地基的地震液化问题。结果表明,将Finn本构模型应用于砂土液化分析,可以较好地给出地震作用过程中孔隙水压力和有效应力变化的规律。     

基于组合权重的模糊物元模型在砂土液化势评价中的应用

目前,砂土液化问题仍是岩土工程界重要的研究方向.鉴于砂土液化评价的复杂性及其影响因素的不确定性和模糊性,将熵值理论与模糊集和贴近度相结合,建立了基于熵权与专家打分法所获权重的组合权重的模糊物元模型,有效地避免了权重分配困难的问题.该模型以一种新的方法进行砂土液化势的评价,并且结合实例应用,通过与现场标贯试验对比,进一步验证了该模型的合理、简便和实用.     

砂土液化大变形本构模型及在ABAQUS软件上的实现

基于Yang和E lgam al等人提出的砂土液化大变形本构模型,对该模型的建立过程进行了详细的推导,基于新的嵌套屈服面硬化规则,对原有模型的硬化规则的不连续性做了改进,把该本构模型扩展应用到三维液化大变形的数值分析中,实现了基于ABAQUS大型商用软件计算平台上砂土液化大变形的计算子程序的开发,基于该计算平台和开发的本构模型,对动三轴试验体系中砂土试样的液化过程进行了数值试验分析,给出了试验过程中试样的竖向动位移、整体竖向应力应变关系滞回曲线和动孔压时程曲线的数值计算结果。文中初步验证了该模型在ABAQUS上开发的子程序的可靠性和数值计算模型的可行性,模型的可靠性及其子程序的稳定性还需通过试验结果和数值计算结果的对比分析与进一步验证。     

A numerical model for dynamic soil liquefaction analysis

This paper presents an effective stress-based numerical model, which can be used to obtain pore pressure build up and consequent loss of soil strength due to earthquake-induced shaking. The main advantage of the new method is that it needs few model parameters compared to many existing effective stress-based ground response analysis methods. The pore pressure generation is calculated using the equivalent cycle pore pressure model developed by Seed et al. [J Geotech Engng Div, ASCE 102 (1976) 323] but the equations are used in a different manner. Pore pressure generation calculated by the new method and the equivalent cycle method for different load patterns shows that the new method can predict pore pressures which are in better agreement with experimental data, irrespective of the loading pattern. The equivalent cycle method predicts results in agreement with experimental data only when the loading pattern is highly irregular, and tends to under-predict pore pressure ratios for other loading patterns. To demonstrate the ability of the new method in simulating earthquake-induced site response and liquefaction-related ground deformations, the Kobe, 1995 earthquake has been analysed. The results obtained from the new analysis agree reasonably well with recorded accelerations and lateral ground displacements at Port Island, Kobe.     

A cyclic UH model for sand

A simple cyclic elastoplastic constitutive model for sand is proposed based on the UH model for overconsolidated clay. The proposed model has the following features. First, in order to describe the stress-induced anisotropy in sand, a rotational hardening rule is introduced for the evolution of the yield surface axis with development of plastic deviatoric strain in the principal stress space. Second, the relationship between the rotational axis and stress-induced anisotropy is modeled by introducing the slope of rotational axis into the yield function. The fl atness of the yield surface can be determined by the slope of rotational axis. Finally, a revised unifi ed hardening parameter is proposed to incorporate the stress-induced anisotropy. The model capability in describing the cyclic response of sand is verifi ed by comparing the simulations with available test results.     

Pre-shear effect on liquefaction resistance of a Fujian sand

Pre-shear history has been shown to be a critical factor in the liquefaction resistance of sand. By contrast to prior experimental studies in which triaxial shear tests were used to examine the effects of pre-shear on the liquefaction resistance of sand, hollow cylinder torsional shear tests were used in this study to avoid the influence of the inherent anisotropy that is inevitably produced during the sample preparation process because of gravitational deposition. A series of cyclic undrained shear tests were performed on sand samples that had experienced medium to large pre-shear loading. The test results showed that the liquefaction resistance of sand can be greatly reduced by its pre-shear history, and a pre-shear strain within the range from 0.1% to 5% can cause sand to be more prone to liquefaction. During the cyclic shear tests, the samples that had experienced pre-shear loading exhibited different behaviors when cyclic shear loading started in different directions, i.e., the clockwise direction and the counterclockwise direction. If the cyclic loading started in the identical direction as the pre-shear loading, then the mean effective stress of the sand was almost unchanged during the first half of the loading cycle; if the cyclic loading started in the direction opposite to that of the pre-shear loading, then the mean effective stress decreased significantly during the first half of the loading cycle. However, this anisotropic behavior was only remarkable during the first loading cycle. From the second cycle onward, the speeds of the decrease in the mean effective stresses in the two types of shear tests became similar.     

基于减法聚类模糊神经网络的砂土液化势判别

砂土地震液化问题是岩土地震工程学的重要研究课题之一。在分析模糊神经网络原理的基础上,利用减法聚类算法对自适应模糊推理系统进行优化,并建立了砂土地震液化的模糊神经网络模型。然后,将该模型用于实际工程的砂土液化判别中,并与传统砂土液化判别方法结果进行对比。判别结果表明:文中建立的模糊神经网络模型具有较强的学习功能,用于砂土地震液化判别中是可行的和有效的。     

基于液化流动模型的尾矿坝地震响应分析

针对传统等效线性法对尾矿坝地震响应分析结果与实际情况存在一定的出入的问题,通过振动台模型试验和数值计算结果比对验证数值计算方法的可行性,运用考虑液化后尾矿砂流动特性的孔压上升模型,基于完全非线性动力分析方法,对尾矿坝地震响应进行分析计算,通过对坝顶位移和加速度的监测结果进行分析,得出坝顶的加速度放大系数以及坝顶位移最大值;并由坝体水平方向位移云图、竖直方向位移云图及位移矢量图,判断可能发生局部滑坡的位置;对坝体内液化情况进行判定,得到尾矿坝的地震响应规律,为尾矿坝抗震设计提供一定的理论参考.     

Non-plastic silty sand liquefaction,screening, and remediation

Assessing liquefaction potential, in situ screening using cone penetration resistance, and liquefaction-remediation of non-plastic silty soils are difficult problems. Presence of silt particles among the sand grains in silty soils alter the moduli, shear strength, and flow characteristics of silty soils compared to clean host sand at the same global void ratio. Cyclic resistance (CRR) and normalized cone penetration resistance (q_c1N) are each affected by silt content in a different way. Therefore, a unique correlation between cyclic resistance and cone resistance is not possible for sands and silty sands. Likewise, the response of silty soils subjected to traditional deep dynamic compaction (DC) and vibro-stone column (SC) densification techniques is influenced by the presence of silt particles, compared to the response in sand. Silty soils require drainage-modifications to make them amenable for dynamic densification techniques. The first part of this paper addresses the effects of silt content on cyclic resistance CRR, hydraulic conductivity k, and coefficient of consolidation C_v of silty soils compared to clean sand. The second part of the paper assesses the effectiveness of equivalent intergranular void ratio (e_c)_eq concept to approximately account for the effects of silt content on CRR. The third part of the paper explores the combined effects of silt content (viz effects of (e_c)_eq, k, and C_v) on q_c1N using laboratory model cone tests and preliminary numerical simulation experiments. A possible inter-relationship between q_c1N, CRR, accommodating the different degrees of influence of (e_c)_eq, k, and C_v on q_c1N and CRR, is discussed. The fourth part of the paper focuses on the detrimental effects of silt content on the effectiveness of DC and SC techniques to densify silty soils for liquefaction-mitigation. Finally, the effectiveness of supplemental wick drains to aid drainage and facilitate densification and liquefaction mitigation of silty sands using DC and SC techniques is discussed.