Modeling of Oblique Load Test on Batter Pile Group Based on Support Vector Machines and Gaussian Regression

This paper evaluates the potential of two machine learning approaches i.e. Support vector machine (SVR) and Gaussian processes (GP) regression to model the oblique load capacity of batter pile groups. Linear regression was used to compare the performance of both SVR and GP based regression approaches to model the oblique load. Data set used consists of 147 samples obtained from the laboratory experiments. Out of the total sample size, 105 randomly selected samples were used for training whereas remaining 42 were used for testing the models. Input data set consist of angle of oblique load, pile length, sand relative density, number of vertical piles, number of batter piles where as oblique load was considered as output. Two kernel functions i.e. Polynomial and radial based kernel function were used with both SVR and GP regression. A comparison of results suggest that radial basis function based SVR approach works well in comparison to GP and linear regression based approaches and it could successfully be employed in modelling the oblique load capacity of batter pile groups. Parametric analysis and sensitivity analysis suggest that loading angle, pile length and number of batter pile were important in prediction of oblique load.     

滑坡灾害空间预测支持向量机模型及其应用

随着GIS技术在滑坡灾害空间预测研究中的广泛应用,滑坡灾害空间预测模型成为研究的热点问题。在总结滑坡灾害空间预测研究现状的基础上,简要介绍了两类和单类支持向量机的基本原理。以香港自然滑坡空间预测为例,采用两类和单类支持向量机进行滑坡灾害空间预测,并与Logistic回归模型进行了比较。结果表明,两类支持向量机模型优于Logistic回归模型,而Logistic回归模型优于单类支持向量机模型。     

Extreme Learning Machine Based Modeling of Resilient Modulus of Subgrade Soils

This paper investigates the potential of two variants of extreme learning machine based regression approaches in predicting the resilient modulus of cohesive soils. Support vector regression was used to compare the performance of the proposed extreme learning machine based regression approaches. The dataset used in this study was derived from literature and consists of 9 input parameters with a total of 891 cases. For testing, two methods i.e. train/test and tenfold cross validation was used. In case of train and test methods, a total of 594 randomly selected cases were used to train different algorithms and the remaining 297 data were used to test the created models. Correlation coefficient value of 0.991 (root mean square error = 3.47 MPa) was achieved by polynomial kernel based extreme learning machine in comparison to 0.990 and 0.990 (root mean square error = 4.790 and 4.290 MPa) by simple extreme learning machine and radial basis kernel function based support vector regression respectively with test dataset. Comparisons of results with tenfold cross validation also suggest that polynomial kernel based extreme learning machine works well in terms of root mean square error and computational cost with the used dataset. Sensitivity analysis suggests the importance of confining stress and deviator stress in predicting the resilient modulus when using with polynomial kernel based extreme learning machine modeling approach.     

滑坡位移的多模态支持向量机模型预测

将支持向量机（support vector machine,SVM）方法与信号分析中的经验模态分解（empirical mode decomposition, EMD）方法相匹配,提出了一种通过多模态支持向量机函数回归分析建模预测滑坡位移的理论方法。以边坡位移历史观测数据为基础,应用EMD方法获得滑坡形成过程中位移演化的几个特征时间模态,构成了多模态信息统计学习样本,确定了边坡位移演化的自适应多尺度变化信息。对应于每个经验模态的位移变化信息,引入了多模态SVM建模方法,然后合成不同经验模态下边坡位移的计算结果,得到滑坡位移的预测值。以卧龙寺新滑坡和新滩滑坡的监测数据为基础的理论预测结果表明,与采用遗传算法的神经网络方法的预测结果相比,支持向量机经验模态方法具有更强的预测能力,理论预测结果与实际监测值具有很好的一致性     

基于改进的SVM技术和高光谱遥感的标准矿物定量计算

基于支持向量机(SVM)统计理论,并对其从核函数构造方面进行改进,通过主成分分析、包络线去除、光谱导数变换等对原始Hyperion高光谱数据进行降维、变换与特征提取,分析比较了这些变换后不同的回归效果,并将其应用在内蒙古霍林郭勒地区岩石中氧化物质量分数的反演中。同时,鉴于某些重要矿物本身并没有明显的特征光谱曲线,提出一种新的矿物定量方法。首先,基于高光谱遥感数据,利用改进的SVM回归技术反演矿物中的化学成分,然后通过标准矿物计算（CIPW）推导岩石中标准矿物的质量分数。研究结果表明：基于改进核函数后的SVM回归精度有所提高,其中导数变换后的反演精度达74.87%,比原始光谱反演精度提高了4.11%。CIPW应用于高光谱遥感地质填图效果良好,为岩性鉴定和评价提供了科学依据。     

Prediction of Backbreak in Open-Pit Blasting Operations Using the Machine Learning Method

Backbreak is an undesirable phenomenon in blasting operations. It can cause instability of mine walls, falling down of machinery, improper fragmentation, reduced efficiency of drilling, etc. The existence of various effective parameters and their unknown relationships are the main reasons for inaccuracy of the empirical models. Presently, the application of new approaches such as artificial intelligence is highly recommended. In this paper, an attempt has been made to predict backbreak in blasting operations of Soungun iron mine, Iran, incorporating rock properties and blast design parameters using the support vector machine (SVM) method. To investigate the suitability of this approach, the predictions by SVM have been compared with multivariate regression analysis (MVRA). The coefficient of determination (CoD) and the mean absolute error (MAE) were taken as performance measures. It was found that the CoD between measured and predicted backbreak was 0.987 and 0.89 by SVM and MVRA, respectively, whereas the MAE was 0.29 and 1.07 by SVM and MVRA, respectively.     

Predicting discharge coefficient of compound broad-crested weir by using genetic programming (GP) and artificial neural network (ANN) techniques

Compound broad-crested weir is a typical hydraulic structure that provides flow control and measurements at different flow depths. Compound broad-crested weir mainly consists of two sections; first, relatively small inner rectangular section for measuring low flows, and a wide rectangular section at higher flow depths. In this paper, series of laboratory experiments was performed to investigate the potential effects of length of crest in flow direction, and step height of broad-crested weir of rectangular compound cross-section on the discharge coefficient. For this purpose, 15 different physical models of broad-crested weirs with rectangular compound cross-sections were tested for a wide range of discharge values. The results of examination for computing discharge coefficient were yielded by using multiple regression equations based on the dimensional analysis. Then, the results obtained were also compared with genetic programming (GP) and artificial neural network (ANN) techniques to investigate the applicability, ability, and accuracy of these procedures. Comparison of results from the GP and ANN procedures clearly indicates that the ANN technique is less efficient in comparison with the GP algorithm, for the determination of discharge coefficient. To examine the accuracy of the results yielded from the GP and ANN procedures, two performance indicators (determination coefficient (R ²) and root mean square error (RMSE)) were used. The comparison test of results clearly shows that the implementation of GP technique sound satisfactory regarding the performance indicators (R ²?=?0.952 and RMSE?=?0.065) with less deviation from the numerical values.     

A distribution-free analysis of magnitude-intensity relationships: an application to the Mediterranean region

A statistical analysis of the relations between macroseismic intensity and magnitude is presented. The examined data set contains earthquakes characterized by epicentral or maximum intensity ≥ VI which occurred in the Mediterranean region. As a first step, an empirical magnitude-intensity relationship has been determined by using the whole data set. Then, differences between experimental magnitude values and the ones expected on the basis of the empirical relationship have been correlated with some features related both to physical and data sources characteristics. On this basis, a distribution-free statistical approach has been developed to attempt a regionalization of the examined area, able to locally optimize the performances of magnitude-intensity relations. However, the results showed that data relative to larger events (intensity ≥ VII) are not sufficient to perform any reliable zonation of the area. Thus, well-constrained relationships determined for the whole Mediterranean region should be preferred to ill-defined local ones. Concerning smaller earthquakes (intensity VI), the analysis suggests that an efficient zonation could only be obtained if medium-scale variations (lower than 200 Km) are taken into account.     

Prediction of blast-induced air overpressure using support vector machine

Prediction of blast-induced air overpressure (AOP) is very complicated and intricate due to the number of influencing parameters affecting air wave propagation. In this paper, an attempt has been made to predict the blast-induced AOP by support vector machine (SVM) using maximum charge per delay and distance from blast-face to monitoring station of AOP. To investigate the suitability of this approach, SVM predictions are compared with a generalized predictor equation. Seventy-five air blasts were monitored at different locations around three mines. AOP data sets of two limestone mines are taken for the training and testing of the SVM network as well as to determine site constants for generalized equation. The remaining mine data sets are used for the validation and comparison of AOP.     

SVM在地下工程可靠性分析中的应用

将支持向量机应用到地下工程可靠性分析中,通过将支持向量机分别与一阶二次矩和蒙特卡洛结合,提出了基于支持向量机的可靠性分析方法,利用数值模拟构造学习样本,通过支持向量机学习,建立变形与随机变量之间映射关系的支持向量机表达,进而实现隧道极限状态函数及其偏导数的显式表达,从而计算隧道的可靠性指标。该方法避免了传统可靠性分析的缺点。算例分析结果表明,该方法计算效率高、结果可靠,对含有大量随机变量的复杂岩土工程可靠性分析具有很大的潜力,具有广泛的应用前景和工程价值。     

Attenuation relations of peak ground acceleration and velocity in the Southern Dead Sea Transform region

Using the recorded earthquake strong ground motion, the attenuation of peak ground acceleration (PGA) and peak ground velocity (PGV) are derived in the southern Dead Sea Transform region. The expected values of strong motion parameters from future earthquakes are estimated from attenuation equations, which are determined by regression analysis on real accelerograms. In this study, the method of Joyner and Boor [Bull Seismol Soc Am 71(6):2011–2038, 1981] was selected to produce the attenuation model for the southern Dead Sea Transform region. The dataset for PGA consists of 57 recordings from 30 earthquakes and for PGV 26 recordings from 19 earthquakes. The attenuation relations developed in this study are proposed as replacement for former probabilistic relations that have been used for a variety of earthquake engineering applications. The comparison between the derived PGA relations from this study with the former relations clearly shows significant lower values than the other relations.     

Slope reliability analysis using a support vector machine

The first-order second-moment method (FOSM) reliability analysis is commonly used for slope stability analysis. It requires the values and partial derivatives of the performance function with respect to the random variables for the design. Such calculations can be cumbersome when the performance functions are implicit. Implicit performance functions are normally encountered when the slope is geologically complicated and the limit equilibrium method (LEM) is used for the stability analysis.

To address this issue, this paper presents a support vector machine (SVM)-based reliability analysis method which combines the SVM with the FOSM. This method employs the SVM method to approximate the implicit performance functions, thus arriving at SVM-based explicit performance functions. The SVM method uses a small set of the actual values of the performance functions obtained via the LEM for complicated slope engineering. Using the SVM model, a large number of values and partial derivatives of the performance functions can be obtained for conventional reliability analysis using the FOSM. Examples are given to illustrate the proposed SVM-based slope reliability analysis. The results show that the proposed approach is applicable to slope reliability analysis which involves implicit performance functions.     

Support vector machines‐based modelling of seismic liquefaction potential

This paper investigates the potential of support vector machines (SVM)‐based classification approach to assess the liquefaction potential from actual standard penetration test (SPT) and cone penetration test (CPT) field data. SVMs are based on statistical learning theory and found to work well in comparison to neural networks in several other applications. Both CPT and SPT field data sets is used with SVMs for predicting the occurrence and non‐occurrence of liquefaction based on different input parameter combination. With SPT and CPT test data sets, highest accuracy of 96 and 97%, respectively, was achieved with SVMs. This suggests that SVMs can effectively be used to model the complex relationship between different soil parameter and the liquefaction potential. Several other combinations of input variable were used to assess the influence of different input parameters on liquefaction potential. Proposed approach suggest that neither normalized cone resistance value with CPT data nor the calculation of standardized SPT value is required with SPT data. Further, SVMs required few user‐defined parameters and provide better performance in comparison to neural network approach. Copyright © 2006 John Wiley & Sons, Ltd.     

Evaluation of liquefaction induced lateral displacements using genetic programming

Determination of liquefaction induced lateral displacements during earthquake is a complex geotechnical engineering problem due to the complex and heterogeneous nature of the soils and the participation of a large number of factors involved. In this paper, a new approach is presented, based on genetic programming (GP), for determination of liquefaction induced lateral spreading. The GP models are trained and validated using a database of SPT-based case histories. Separate models are presented to estimate lateral displacements for free face and for gently sloping ground conditions. It is shown that the GP models are able to learn, with a very high accuracy, the complex relationship between lateral spreading and its contributing factors in the form of a function. The attained function can then be used to generalize the learning to predict liquefaction induced lateral spreading for new cases not used in the construction of the model. The results of the developed GP models are compared with those of a commonly used multi linear regression (MLR) model and the advantages of the proposed GP model over the conventional method are highlighted.     

基于组合核函数的高斯过程边坡角智能设计

高斯过程(GP)是近年来发展迅速的一种全新学习机。与支持向量机(SVM)相比,该方法有着容易实现、超参数可自适应获取及预测输出具有概率意义等优点。结合边坡工程中的边坡角设计,编写了在多种因素影响下边坡角设计的GP程序,为克服单一核函数预测精度和网络泛化能力差的缺点,采用单一核函数相加作为GP的组合核函数,将自动关联性测定参数（ARD）引入其中,建立了关于超参数的GP回归网络模型,使用共轭梯度下降算法导出最优超参数,用ARD超参数进行输入属性相关性分析和特征选取,并以此网络对测试样本进行学习预测,结合支持向量回归方法给出了在回归问题上的应用和对比分析。结果表明：在边坡角智能设计应用中,采用组合核函数的GPR网络ARD参数具有明确的物理意义,预测回归性能优于SVM,且预测输出的概率解释能更好的体现预测值的代表性,为边坡角设计开辟新径。     

Gaussian Processes for history-matching: application to an unconventional gas reservoir

The process of reservoir history-matching is a costly task. Many available history-matching algorithms either fail to perform such a task or they require a large number of simulation runs. To overcome such struggles, we apply the Gaussian Process (GP) modeling technique to approximate the costly objective functions and to expedite finding the global optima. A GP model is a proxy, which is employed to model the input-output relationships by assuming a multi-Gaussian distribution on the output values. An infill criterion is used in conjunction with a GP model to help sequentially add the samples with potentially lower outputs. The IC fault model is used to compare the efficiency of GP-based optimization method with other typical optimization methods for minimizing the objective function. In this paper, we present the applicability of using a GP modeling approach for reservoir history-matching problems, which is exemplified by numerical analysis of production data from a horizontal multi-stage fractured tight gas condensate well. The results for the case that is studied here show a quick convergence to the lowest objective values in less than 100 simulations for this 20-dimensional problem. This amounts to an almost 10 times faster performance compared to the Differential Evolution (DE) algorithm that is also known to be a powerful optimization technique. The sensitivities are conducted to explain the performance of the GP-based optimization technique with various correlation functions.     

A case study using support vector machines, neural networks and logistic regression in a GIS to identify wells contaminated with nitrate-N

Accurate and inexpensive identification of potentially contaminated wells is critical for water resources protection and management. The objectives of this study are to 1) assess the suitability of approximation tools such as neural networks (NN) and support vector machines (SVM) integrated in a geographic information system (GIS) for identifying contaminated wells and 2) use logistic regression and feature selection methods to identify significant variables for transporting contaminants in and through the soil profile to the groundwater. Fourteen GIS derived soil hydrogeologic and landuse parameters were used as initial inputs in this study. Well water quality data (nitrate-N) from 6,917 wells provided by Florida Department of Environmental Protection (USA) were used as an output target class. The use of the logistic regression and feature selection methods reduced the number of input variables to nine. Receiver operating characteristics (ROC) curves were used for evaluation of these approximation tools. Results showed superior performance with the NN as compared to SVM especially on training data while testing results were comparable. Feature selection did not improve accuracy; however, it helped increase the sensitivity or true positive rate (TPR). Thus, a higher TPR was obtainable with fewer variables.     

Disaster prediction model based on support vector machine for regression and improved differential evolution

The kernel parameters setting of SVM influences prediction precision. The hybrid model based on SVM for regression and improved differential evolution is proposed to enhance the prediction precision. The improved differential evolution is used to optimize the kernel parameters. The improved differential evolution algorithm employs two trial vector generation strategies and two control parameter settings. The first-generation strategy is with best solution, and the second strategy is without best solution. Three categories of disasters time series including flood, drought and storm from Ministry of agriculture of China are used to verify the validity of the proposed model. Compared with the grid SVM and other models, the proposed hybrid model improves the prediction precision of SVM.     

Comparison of machine-learning techniques for landslide susceptibility mapping using two-level random sampling (2LRS) in Alakir catchment area,Antalya, Turkey

The aim of this study is to make a comparison of the performances of two machine-learning algorithms that support vector machine (SVM) and random forest (RF) for landslide susceptibility mapping. The study makes use of a sampling strategy called two-level random sampling (2LRS). During landslide susceptibility mapping, training and testing samples must be collected from different landslide seed cells, which are then put through a fully independent sampling using the 2LRS algorithm. This approach requires fewer samples for the improvement of the computation time of both machine-learning classifications. The proposed approach was tested in the Alakir catchment area (Western Antalya, Turkey) which features numerous active deep-seated rotational landslides. In order to compare the performance of the machine-learning algorithms, three random sets were generated for SVM and three random sets generated for 10, 100, 1000 and 10,000-tree size RF. A total of 15 models were generated for comparison, and their spatial performances were performed by the area under the receiver-operating characteristic curves, which ranged between 0.82 and 0.87. The highest and lowest performances were recorded from two models in SVM and two models from the 1000-tree and 10,000-tree sized RF, respectively. These results were confirmed the landslide happened just after producing the susceptibility maps in the field.     

Is spatial autocorrelation introducing biases in the apparent accuracy of paleoclimatic reconstructions?

We address the issue of spatial autocorrelation, an occurrence that may introduce biases in the evaluation of the performance of transfer functions, by using two fundamentally different approaches, one based on calibration (weighted averaging partial least squares regressions; WA-PLS) and the other based on similarities (modern analogue technique; MAT). The tests were made after spatial standardization of 700 North Atlantic surface data points, which include 29 dinocyst taxa and 4 climate parameters. The evaluation of transfer function performance was made by defining a verification dataset that was gradually isolated from the calibration or comparison datasets. Although strong spatial autocorrelation characterizes the original climate parameter distribution, the results show that the spatial structure of data has relatively low effect on the calculation of the error of prediction. They also show that the performances of MAT are generally better than those of WA-PLS, with lower error of prediction. The better performance of MAT in the present study can be explained by the non-modal distribution of salinity and temperature in the studied marine environments, which is not appropriate for the application of WA-PLS. The two methods yield equivalent results about the spatial structure of the residuals based on empirical semi-variograms. The analyses we performed include the comparison of reconstructions based on original raw data and gridded data. Results suggest that the gridding of the reference database may reduce the noise and thus improve the performance of the techniques.