Prediction of landslide displacement with an ensemble-based extreme learning machine and copula models

Research on the dynamics of landslide displacement forms the basis for landslide hazard prevention. This paper proposes a novel data-driven approach to monitor and predict the landslide displacement. In the first part, autoregressive moving average time series models are constructed to analyze the autocorrelation of landslide triggering factors. A linear ensemble-based extreme learning machine using the least absolute shrinkage and selection operator is applied in predicting the displacement of landslides. Five benchmarking data-driven models, the support vector machine, neural network, random forest, k-nearest neighbor, and the classical extreme learning machine, are considered as baseline models for validating the ensemble-based extreme learning machines. Numerical experiments demonstrated that the proposed prediction model produces the smallest prediction errors among all the algorithms tested. In the second part, parametric copula models are fitted on the predicted displacement, to investigate the relationship between the triggering factors and landslide displacement values. The Gumbel-Hougaard copula model performs best, which indicates strong upper tail correlation between the triggering factors and displacement values. Thresholds for the triggering factors can be obtained by monitoring the landslide moving patterns with large displacement values. The effectiveness and utility of the proposed data-driven approach have been confirmed with the landslide case study in the region of the Three Gorges Reservoir.     

Piles shaft capacity from CPT and CPTu data by polynomial neural networks and genetic algorithms

Cone penetration test (CPT) is one of the most common in situ tests which is used for pile design because it can be realized as a model pile. The measured cone resistance (q_c) and sleeve friction (f_s) usually are employed for estimation of pile unit toe and shaft resistances, respectively. Thirty three pile case histories have been compiled including static loading tests performed in uplift, or in push with separation of shaft and toe resistances at sites which comprise CPT or CPTu sounding. Group method of data handling (GMDH) type neural networks optimized using genetic algorithms (GAs) are used to model the effects of effective cone point resistance (q_E) and cone sleeve friction (f_s) as input parameters on pile unit shaft resistance, applying some experimentally obtained training and test data. Sensitivity analysis of the obtained model has been carried out to study the influence of input parameters on model output. Some graphs have been derived from sensitivity analysis to estimate pile unit shaft resistance based on q_E and f_s. The performance of the proposed method has been compared with the other CPT and CPTu direct methods and referenced to measured piles shaft capacity. The results demonstrate that appreciable improvement in prediction of pile shaft capacity has been achieved.     

Prediction of pile shaft resistance using cone penetration tests (CPTs)

Accurately predicting pile shaft resistance when designing pile foundations is necessary for ensuring appropriate structural and serviceability performance. The scope of this research includes four main components: (I) compiling shaft resistance datasets obtained from the published literature; (II) developing two artificial neural network (ANN) and non-linear multi regression models for predicting pile shaft resistance using cone penetration test (CPT) results; (III) investigating the influence of input parameters on the resulting shaft friction and their degrees of importance; and (IV) assessing the relative accuracies of the presented models using a number of traditional methods. It is quantitatively demonstrated that the ANN and non-linear multiple regression models proposed in the current study out perform the traditional methods and can be used by engineers to accurately predict pile shaft resistance.     

Uplift capacity of single piles: predictions and performance

The paper pertains to the development of a simple semi-empirical model for predicting the uplift capacity of piles embedded in sand. Various pile and soil parameters such as length (L), diameter (d) of the pile and angle of friction (ϕ), soil–pile friction angle (δ) and unit weight (γ) of the soil which have direct influence on the uplift capacity of the pile are incorporated in the analysis. A comparative assessment of the ultimate uplift capacity of piles predicted by using the proposed theory and some of the available theories are made with respect to each other and with reference to the measured values obtained from model tests in the laboratory. For this purpose experimental data have been collected from the literature and also from model tests conducted as a part of the present investigation. The study shows the proposed model has an excellent potential in predicting the uplift capacity of piles embedded in sand that are consistent with model pile test results.     

Prediction of pile settlement using artificial neural networks based on standard penetration test data

In recent years artificial neural networks (ANNs) have been applied to many geotechnical engineering problems with some degree of success. With respect to the design of pile foundations, accurate prediction of pile settlement is necessary to ensure appropriate structural and serviceability performance. In this paper, an ANN model is developed for predicting pile settlement based on standard penetration test (SPT) data. Approximately 1000 data sets, obtained from the published literature, are used to develop the ANN model. In addition, the paper discusses the choice of input and internal network parameters which were examined to obtain the optimum model. Finally, the paper compares the predictions obtained by the ANN with those given by a number of traditional methods. It is demonstrated that the ANN model outperforms the traditional methods and provides accurate pile settlement predictions.     

软岩地基中大直径布袋桩抗拔试验研究

西北地区深大基础工程日益增多,兼顾基础抗浮和耐久性问题的研究空白,借助西宁火车站综合改造工程,引入大直径布袋桩技术,有效解决了基础抗浮和耐久性问题;选择6根试桩进行了现场单桩抗拔载荷试验,最大加载量为9 060 kN;运用MATLAB软件分别拟合出3种抗拔极限承载力预测函数模型的曲线,同时运用PLAXIS软件对不同等级荷载桩-土位移进行模拟,并与实测的荷载-位移曲线对比分析。研究发现：双曲线和幂函数模型较适合此类抗拔桩极限承载力预测;本地区类似地基预测大直径缓变形抗拔桩极限荷载所需的极限位移标准应由0.030D减小为0.025D;仅根据土层的物理力学特征确定抗拔桩桩周土的极限摩阻力不够完善,至少还要考虑埋深不同对具有相似物理力学特征土层性质的影响。     

联合多种数据驱动建模方法的滑坡位移预测研究

边坡位移是滑坡演化的宏观体现,分析并预测滑坡位移发展态势对于防灾减灾具有重要意义。由于滑坡位移曲线具有明显的非线性特征,单一模型往往难以刻画其非线性与复杂性。为发展一种普遍适用于滑坡位移的预测方法,提出了一种联合多种数据驱动模型的新方法。该方法根据时间序列分析理论,将滑坡位移序列分解为趋势项和周期项,趋势项采用并联型灰色神经网络处理,周期项则采用人工蜂群算法（ABC）优化后的极限学习机模型（ELM）处理,从而充分应用各种模型的优点。以三峡库区白水河和八字门滑坡为例,对位移数据进行分析处理后,灰色神经网络模型预测其趋势性位移,改进后的极限学习机模型对周期性位移进行训练及预测。结果表明:在预测精度上,优化后的极限学习机模型准确度高于极限学习机模型及小波神经网络等方法,提出的灰色神经网络与ABC-ELM的组合模型可作为实际工程的一个参考。     

Modelling load–displacement response of driven piles in cohesionless soils under tensile loading

A new methodology for deriving the uplift load–displacement response of long driven piles in cohesionless soils is proposed. This method accounts for the effects of the friction fatigue processes during pile driving and the existence of locked-in residual stresses at the end of pile driving before commencing the pile load test. A hyperbolic formulation is utilized to simulate the nonlinear load transfer curves (the so-called t–z curves). The utility of this technique is demonstrated for a field pullout load test on a driven pile in sand. Predicted and measured load–displacement curves showed good agreement, indicating that this approach yields reasonable results as long as representative input parameters are employed.     

Least square support vector machine and relevance vector machine for evaluating seismic liquefaction potential using SPT

The determination of liquefaction potential of soil is an imperative task in earthquake geotechnical engineering. The current research aims at proposing least square support vector machine (LSSVM) and relevance vector machine (RVM) as novel classification techniques for the determination of liquefaction potential of soil from actual standard penetration test (SPT) data. The LSSVM is a statistical learning method that has a self-contained basis of statistical learning theory and excellent learning performance. RVM is based on a Bayesian formulation. It can generalize well and provide inferences at low computational cost. Both models give probabilistic output. A comparative study has been also done between developed two models and artificial neural network model. The study shows that RVM is the best model for the prediction of liquefaction potential of soil is based on SPT data.     

Prediction of ultimate axial load-carrying capacity of piles using a support vector machine based on CPT data

The support vector machine (SVM) is a relatively new artificial intelligence technique which is increasingly being applied to geotechnical problems and is yielding encouraging results. In this paper SVM models are developed for predicting the ultimate axial load-carrying capacity of piles based on cone penetration test (CPT) data. A data set of 108 samples is used to develop the SVM models. These data were obtained from the literature containing pile load tests and each sample contains information regarding pile geometry, full-scale static pile load tests and CPT results. Moreover, a sensitivity analysis is carried out to examine the relative significance of each input variable with respect to ultimate strength prediction. Finally, a statistical analysis is conducted to make comparisons between predictions obtained from the SVM models and three traditional CPT-based methods for determining pile capacity. The comparison confirms that the SVM models developed in this paper outperform the traditional methods.     

Longitudinal vibration of a pile embedded in layered soil considering the transverse inertia effect of pile

The dynamic response of a viscoelastic bearing pile embedded in multilayered soil is theoretically investigated considering the transverse inertia effect of the pile. The soil layers surrounding the pile are modeled as a set of viscoelastic continuous media in three-dimensional axisymmetric space, and a simplified model, i.e., the distributed Voigt model, is proposed to simulate the dynamic interactions of the adjacent soil layers. Meanwhile, the pile is assumed to be a Rayleigh–Love rod with material damping and can be divided into several pile segments allowing for soil layers and pile defects. Both the vertical and radial displacement continuity conditions at the soil–pile interface are taken into account. The potential function decomposition method and the variable separation method are introduced to solve the governing equations of soil vibration in which the vertical and radial displacement components are coupled. On this basis, the impedance function at the top of the pile segment is derived by invoking the force and displacement continuity conditions at the soil–pile interface as well as the bottom of pile segment. The impedance function at the pile head is then obtained by means of the impedance function transfer method. By means of the inverse Fourier transform and convolution theorem, the velocity response in the time domain can also be obtained. The reasonableness of the assumptions of the soil-layer interactions have been verified by comparing the present solutions with two published solutions and a set of well-documented measured pile test data. A parametric analysis is then conducted using the present solutions to investigate the influence of the transverse inertia effect on the dynamic response of an intact pile and a defective pile for different design parameters of the soil–pile system.     

Numerical ANFIS-Based Formulation for Prediction of the Ultimate Axial Load Bearing Capacity of Piles Through CPT Data

This study explores the potential of adaptive neuro-fuzzy inference systems (ANFIS) for prediction of the ultimate axial load bearing capacity of piles (P_u) using cone penetration test (CPT) data. In this regard, a reliable previously published database composed of 108 datasets was selected to develop ANFIS models. The collected database contains information regarding pile geometry, material, installation, full-scale static pile load test and CPT results for each sample. Reviewing the literature, several common and uncommon variables have been considered for direct or indirect estimation of P_u based on static pile load test, cone penetration test data or other in situ or laboratory testing methods. In present study, the pile shaft and tip area, the average cone tip resistance along the embedded length of the pile, the average cone tip resistance over influence zone and the average sleeve friction along the embedded length of the pile which are obtained from CPT data are considered as independent input variables where the output variable is P_u for the ANFIS model development. Besides, a notable criticism about ANFIS as a prediction tool is that it does not provide practical prediction equations. To tackle this issue, the obtained optimal ANFIS model is represented as a tractable equation which can be used via spread sheet software or hand calculations to provide precise predictions of P_u with the calculated correlation coefficient of 0.96 between predicted and experimental values for all of the data in this study. Considering several criteria, it is represented that the proposed model is able to estimate the output with a high degree of accuracy as compared to those results obtained by some direct CPT-based methods in the literature. Furthermore, in order to assess the capability of the proposed model from geotechnical engineering viewpoints, sensitivity and parametric analyses are done.     

A Novel Methodology to Classify Soil Liquefaction Using Deep Learning

In this research, deep learning (DL) model is proposed to classify the soil reliability for liquefaction. The applicability of the DL model is tested in comparison with emotional backpropagation neural network (EmBP). The database encompassing cone penetration test of Chi–Chi earthquake. This study uses cone resistance (q_c) and peck ground acceleration as inputs for prediction of liquefaction susceptibility of soil. The performance of developed models has been assessed by using various parameters (receiver operating characteristic, sensitivity, specificity, Phi correlation coefficient, Precision–Recall F measure). The performance of DL is excellent. Consistent results obtained from the proposed deep learning model, compared to the EmBP, indicate the robustness of the methodology used in this study. In addition, both the developed model was also tested on global earthquake data. During validation on global data, both the models shows good results based on fitness parameters. The developed classification models a simple, but also efficient decision-making tool in engineering design to quantitatively assess the liquefaction potential. The finding of this paper can be further used to capture the relationship between soil and earthquake parameters.

     

Numerical analyses of uplift behavior of pile foundation for transmission line structure in frozen soil regions北大核心CSCD

Pile foundation is one of the most commonly used and suitable foundations to support transmission line structure, especially in seasonally frozen soil regions and permafrost regions. Axial compression is the controlling condition in the design of foundations for such structures as bridges and buildings, while uplift and overturning will control the design of transmission line structure foundations. This paper presents an extensive overview of previous studies including experimental (e. g., laboratory model test and full-scale field load test), analytical/theoretical (e. g., limit equilibrium and limit analysis based on plasticity)and numerical(e. g., finite difference and finite element methods). The review indicates that study on the uplift behavior of pile foundation in frozen soil is relatively limited, particularly in the case of combined effect of axial uplift and lateral loading. Interaction between pile and frozen soil and mechanism of load transfer along the pile shaft and around the pile tip still remain unclear. Therefore, this paper implements finite difference analysis within FLAC3D to investigate the behavior of pile foundation in frozen silty clay and gravelly sand under axial uplift behavior and the effect of ground condition and lateral loading on the uplift behavior. Because of the axisymmetric condition of the problem studied, only half of the model is simulated. The chosen domain of the medium is discretized into a set of quadrilateral elements and the pile is discretized by the cylinder element. The interaction between the soil and pile is considered according to interface elements. Mohr-Coulomb criterion is adopted to model the soil behavior (perfectly elastic-plastic), while the pile is simply considered as a rigid body. The soil parameters such as Young’s modulus, cohesion and internal friction angle used for numerical analyses are determined by laboratory tests and estimated according to the empirical correlations with in-situ tests. The present numerical modeling is verified with the results from field loading tests on pile foundations in Qinghai-Tibet ±550 kV transmission line project. On this basis, parametric studies are carried out to uncover the behavior of pile in frozen soil. It is observed that pullout is the dominant failure mechanism of pile and the uplift load-displacement curve clearly exhibits an asymptote, consisting of initially linear elastic, nonlinear transition, and finally linear regions. These results are consistent with the observations in a few previous studies. In addition, larger uplift capacity of pile foundation in freezing period and gravelly sand is gained (about 20%). Lateral loading increases the deflection and therefore, decreases the uplift capacity of pile foundation. For the convenience of using the results obtained in practice, the values of uplift factor for pile foundation in silty clay and gravelly sand are provided. Finally, it should be noted that the method used, and the results obtained in the current work could be useful for engineers and designers, at least providing them some qualitative evidence for pile design in seasonally frozen soil regions and permafrost regions. This is important and necessary to ensure the safety of construction in such regions. Meanwhile, numerical analyses in the current work can be a benchmark example for subsequent research studies. © 2022 Science Press (China).     

极限学习机模型在季冻区深基坑地表沉降预测中的应用

针对传统数据驱动模型存在收敛速度慢、过度拟合等问题,提出了基于极限学习机算法的基坑地表沉降预测方法。结合季冻区地铁车站基坑的特点,提取基坑开挖时间、开挖深度、围护桩顶位移、围护桩内力、支撑轴力及地表温度等特征信息,建立极限学习机回归预测模型,选用实例数据进行算例分析,并将其与传统回归预测模型进行对比,实验结果表明,极限学习机模型收敛速度快,泛化能力强,其预测精度优于传统预测模型,且在学习速度方面优势明显,对深基坑安全监控有一定的实用价值。     

扩底抗拔桩桩端阻力的群桩效应研究

通过颗粒流数值模拟,从桩端阻力随上拔位移的发展与桩端周围土体颗粒位移表现等角度,研究了扩底抗拔桩端阻力的群桩效应问题。研究比较了单桩（墩）与群桩（墩）的抗拔性状以及不同墩距下中心墩与边墩阻力随上拔位移发展的情况。研究表明,在归一化上拔量 0.1时,单桩（墩）与群桩（墩）的上拔特性无明显差别;此后,随着上拔位移的发展,单桩（墩）的上拔端阻力要大于群桩中的桩（墩）的端阻力,桩（墩）周围土体颗粒的相互影响开始显现。在归一化上拔量 0.5的情况下,群桩（墩）中中心桩（墩）的端阻力要略大于边桩（墩）。在归一化上拔量 0.5的情况下,群桩中边桩的桩端阻力较中心桩的要大,而群墩中边墩的墩端阻力较中心墩要小,体现了桩身侧限对抗拔桩群中端阻力发挥的影响。随着墩距的增大,在较大的位移量以后群墩才与单墩的受力有显著差别。     

抗拔桩轴力分布函数和位移分布函数间的关系模型

为了从荷载传递微分方程导出具有实际意义的非线性解析解,建立一种简单的沿桩长轴力分布函数和位移分布函数之间的关系模型U(z)-?(z)是非常重要的。采用建立经验公式类数学模型的方法,从几何作图法获得的U(z)-?(z)曲线的形状入手,通过对若干个数学模型的试算,确定了指数形式的模型,并对模型中的参数意义和确定进行了讨论,认为参数?,b只与桩顶和桩底位移有关。     

基于时间序列分解和多变量混沌模型的滑坡阶跃式位移预测

三峡库区某些库岸滑坡在强降雨、库水位涨落等诱发因素影响下,其位移时间序列表现出阶跃式变化特征且可能存在混沌特性.但目前常用于滑坡位移预测的混沌模型,均建立在单变量混沌理论的基础之上.且已有的考虑了诱发因素的常规多变量模型,大都采用经验性的方法来选取输入变量;常规多变量模型对滑坡位移序列的非线性特征,及其与诱发因素间的动态响应关系缺乏数学理论上的深入分析.因此,提出一种基于指数平滑法、多变量混沌模型和极限学习机（extreme learing machine,ELM）的滑坡位移组合预测模型.指数平滑多变量混沌ELM模型首先对滑坡累积位移序列的混沌特性进行识别;然后用指数平滑法对累积位移进行预测,得到趋势项位移,并用累积位移减去趋势项位移得到剩余的波动项位移;之后对波动项位移及降雨量、库水位变化量这3个因子进行多变量相空间重构,并用ELM模型对多变量重构后的波动项位移进行预测;最后将预测得到的趋势项和波动项位移值相加,得到最终的累积位移预测值.以三峡库区白水河滑坡ZG93监测点的累积位移作为实例进行分析,并将模型与指数平滑多变量混沌粒子群-支持向量机（PSO-SVM）模型、指数平滑单变量混沌ELM模型作对比.结果表明滑坡位移序列存在混沌特性,模型能有效预测滑坡位移,其预测效果优于对比模型.且本文模型从混沌理论的角度将波动项位移与降雨量、库水位变化量的动态响应关系进行综合分析,更能反映滑坡位移系统演化的物理本质.      

Prediction of load displacement response of single piles under uplift load

Load displacement response and ultimate resistance of piles in sand under uplift load are predicted by load transfer approach. The pile is divided into number of segments and assigned geometrical and material properties according to actual soil pile situation. The shaft resistance is obtained analytically in accordance with existing studies. The proposed method takes into account the length, diameter and relevant surface characteristics of pile and soil properties. The load displacement characteristics and the value of uplift capacity of vertical piles from field test have been predicted. Reasonable agreement has been found out between predicted and observed values of uplift capacity. Load transfer mechanism is capable of predicting the nonlinear variation of load-displacement response of piles.     

How do machine learning techniques help in increasing accuracy of landslide susceptibility maps?

Landslides are abundant in mountainous regions.They are responsible for substantial damages and losses in those areas.The A1 Highway,which is an important road in Algeria,was sometimes constructed in mountainous and/or semi-mountainous areas.Previous studies of landslide susceptibility mapping conducted near this road using statistical and expert methods have yielded ordinary results.In this research,we are interested in how do machine learning techniques help in increasing accuracy of landslide susceptibility maps in the vicinity of the A1 Highway corridor.To do this,an important section at Ain Bouziane(NE,Algeria) is chosen as a case study to evaluate the landslide susceptibility using three different machine learning methods,namely,random forest(RF),support vector machine(SVM),and boosted regression tree(BRT).First,an inventory map and nine input factors were prepared for landslide susceptibility mapping(LSM) analyses.The three models were constructed to find the most susceptible areas to this phenomenon.The results were assessed by calculating the receiver operating characteristic(ROC) curve,the standard error(Std.error),and the confidence interval(CI) at 95%.The RF model reached the highest predictive accuracy(AUC=97.2%) comparatively to the other models.The outcomes of this research proved that the obtained machine learning models had the ability to predict future landslide locations in this important road section.In addition,their application gives an improvement of the accuracy of LSMs near the road corridor.The machine learning models may become an important prediction tool that will identify landslide alleviation actions.