昆仑山地区冻融土导热系数试验测试与预测模型研究

为探究青藏高原工程走廊带昆仑山地区冻融土导热系数基本特征,采用瞬态平面热源法对钻取的349组冻土试样和245组融土试样导热系数进行了测试,分析了五类土导热系数分布特征及天然含水率、干密度与导热系数的偏相关性,并以两者为变量因素建立了经验公式拟合、支持向量回归（SVR）和径向基（RBF）神经网络导热系数预测模型。结果表明:冻融土导热系数整体均呈粗颗粒土大于细颗粒土特征,且冻土和融土导热系数随土性分布规律存在差异;天然含水率、干密度与导热系数均呈正相关性,不同土类偏相关性结果差异明显,典型土导热系数二元经验回归方程表现为非线性拟合结果。对比三种预测模型下各典型土冻融土导热系数预测结果,全风化千枚岩、角砾及砾砂三种预测模型效果整体较佳,粉土的SVR及RBF神经网络预测精度较好;融土导热系数预测效果整体略优于冻土,SVR及RBF神经网络模型下角砾、粉土及全风化千枚岩融土导热系数预测精度较高。综合导热系数模型预测效果和误差结果分析可得,SVR和RBF神经网络模型预测效果显著优于经验方程拟合,后者针对部分土性拟合效果相对较好,可满足一般工程估算需求;SVR和RBF神经网络预测模型针对不同土性导热系数预测效果呈差异性变化,整体预测效果相当,且预测精度更高、应用土性范围更广。     

Prediction of pore-water pressure response to rainfall using support vector regression

Nonlinear complex behavior of pore-water pressure responses to rainfall was modelled using support vector regression (SVR). Pore-water pressure can rise to disturbing levels that may result in slope failure during or after rainfall. Traditionally, monitoring slope pore-water pressure responses to rainfall is tedious and expensive, in that the slope must be instrumented with necessary monitors. Data on rainfall and corresponding responses of pore-water pressure were collected from such a monitoring program at a slope site in Malaysia and used to develop SVR models to predict pore-water pressure fluctuations. Three models, based on their different input configurations, were developed. SVR optimum meta-parameters were obtained using k-fold cross validation and a grid search. Model type 3 was adjudged the best among the models and was used to predict three other points on the slope. For each point, lag intervals of 30 min, 1 h and 2 h were used to make the predictions. The SVR model predictions were compared with predictions made by an artificial neural network model; overall, the SVR model showed slightly better results. Uncertainty quantification analysis was also performed for further model assessment. The uncertainty components were found to be low and tolerable, with d-factor of 0.14 and 74 % of observed data falling within the 95 % confidence bound. The study demonstrated that the SVR model is effective in providing an accurate and quick means of obtaining pore-water pressure response, which may be vital in systems where response information is urgently needed.     

Pre-processing data using wavelet transform and PCA based on support vector regression and gene expression programming for river flow simulation

An accurate estimation of flow using different models is an issue for water resource researchers. In this study, support vector regression (SVR) and gene expression programming (GEP) models in daily and monthly scale were used in order to simulate Gamasiyab River flow in Nahavand, Iran. The results showed that although the performance of models in daily scale was acceptable and the result of SVR model was a little better, their performance in the daily scale was really better than the monthly scale. Therefore, wavelet transform was used and the main signal of every input was decomposed. Then, by using principal component analysis method, important sub-signals were recognized and used as inputs for the SVR and GEP models to produce wavelet-support vector regression (WSVR) and wavelet-gene expression programming. The results showed that the performance of WSVR was better than the SVR in such a way that the combination of SVR with wavelet could improve the determination coefficient of the model up to 3% and 18% for daily and monthly scales, respectively. Totally, it can be said that the combination of wavelet with SVR is a suitable tool for the prediction of Gamasiyab River flow in both daily and monthly scales.     

基于整体经验模态分解和支持向量回归的北京市 PM 2．5预测

为了更好地掌握大气中PM2.5浓度的变化规律,利用EEMD-SVR混合模型对该地区的PM2.5浓度值进行了短期预测。首先,通过采用整体经验模态分解(EEMD)方法分析北京市PM2.5,把原始时间序列分解成多个固有模态函数和趋势项;然后,对各阶固有模态函数进行周期性分析,揭示了北京市PM2.5的周期性变化特点;最后,对经过EEMD分解后的各阶固有模态函数和趋势项用支持向量机回归(SVR)方法进行预测。结果表明,EEMD-SVR混合模型比单一的SVR模型预测精度更高。     

采用支持向量回归从测井曲线定量计算油页岩含油率

获取油页岩含油率是油页岩资源评价的重要步骤,传统从测井曲线计算油页岩含油率多采用回归模型,但存在误差大或过拟合的局限性和弱点.本文尝试结合大数据概念的数据挖掘算法和测井应用知识进行油页岩含油率定量计算,提高含油率计算的精度以及模型的泛化性.利用改进的ΔlogR技术获得DT_s、DEN_s、GR_s作为解释变量.采用数据挖掘算法——支持向量回归进行定量计算油页岩含油率能够大幅提高泛化性和精度,获得模型训练样本R²得分为0.82,测试样本R²得分可达0.70,拟合精度较高.支持向量回归模型比传统回归模型泛化能力更强,能够避免过拟合问题,具有广泛的应用性.     

基于聚类和支持向量回归的短期负荷预测方法

提出了一种基于自组织特征映射(SOFM)的聚类分析和支持向量回归(SVR)的电力系统短期负荷预测方法。该方法首先利用自组织特征映射网络,通过无监督学习策略,对训练样本集进行聚类分析,将其分为若干相似子类;再针对每一子类构造一个支持向量回归(SVR)模型,以对应子类的样本集训练SVR模型。由于聚类后的每一子类的样本具有相似性,同时子类样本数较少,因此,该方法能够缩短训练时间,提高预测精度。基于某电网提供的历史负荷数据进行的不同方法对比实验说明了该方法的有效性。     

Applying Optimized Support Vector Regression Models for Prediction of Tunnel Boring Machine Performance

One of the main factors in the effective application of a tunnel boring machine (TBM) is the ability to accurately estimate the machine performance in order to determine the project costs and schedule. Predicting the TBM performance is a nonlinear and multivariable complex problem. The aim of this study is to predict the performance of TBM using the hybrid of support vector regression (SVR) and the differential evolution algorithm (DE), artificial bee colony algorithm (ABC), and gravitational search algorithm (GSA). The DE, ABC and GSA are combined with the SVR for determining the optimal value of its user defined parameters. The optimization implementation by the DE, ABC and GSA significantly improves the generalization ability of the SVR. The uniaxial compressive strength (UCS), average distance between planes of weakness (DPW), the angle between tunnel axis and the planes of weakness (α), and intact rock brittleness (BI) were considered as the input parameters, while the rate of penetration was the output parameter. The prediction models were applied to the available data given in the literature, and their performance was assessed based on statistical criteria. The results clearly show the superiority of DE when integrated with SVR for optimizing values of its parameters. In addition, the suggested model was compared with the methods previously presented for predicting the TBM penetration rate. The comparative results revealed that the hybrid of DE and SVR yields a robust model which outperforms other models in terms of the higher correlation coefficient and lower mean squared error.     

Support vector regression to predict asphalt mix performance

Material properties are essential in the design and evaluation of pavements. In this paper, the potential of support vector regression (SVR) algorithm is explored to predict the resilient modulus (M_R), which is an essential property in designing and evaluating pavement materials, particularly hot mix asphalt typically used in Oklahoma. SVR is a statistical learning algorithm that is applied to regression problems; in our study, SVR was shown to be superior to the least squares (LS). Compared with the widely used LS method, the results of this study show that SVR significantly reduces the mean‐squared error and improves the correlation coefficient. Copyright © 2008 John Wiley & Sons, Ltd.     

一种基于核学习的储集层渗透率预测新方法

基于核学习的支持向量机,是一种采用结构风险最小化原则代替传统经验风险最小化原则的新型统计学习方法,具有完备的理论基础。这里提出了核学习技术在储集层非均质特性描述中渗透率参数预测的新用途。在复杂地层中,基于支持向量机的智能和自适应模式识别能力而建立了常规测井多参数信息输入的渗透率预测模型,然后对实际油田储集层渗透率进行了预测。与常规线性回归模型预测结果相对比,所提出的方法更易于使用,很少受不确定因素的影响,并具有较强的信息整合能力以及更高的预测准确性和可信度。     

Prediction of swelling pressure of soil using artificial intelligence techniques

The swelling pressure of soil depends upon various soil parameters such as mineralogy, clay content, Atterberg’s limits, dry density, moisture content, initial degree of saturation, etc. along with structural and environmental factors. It is very difficult to model and analyze swelling pressure effectively taking all the above aspects into consideration. Various statistical/empirical methods have been attempted to predict the swelling pressure based on index properties of soil. In this paper, the computational intelligence techniques artificial neural network and support vector machine have been used to develop models based on the set of available experimental results to predict swelling pressure from the inputs; natural moisture content, dry density, liquid limit, plasticity index, and clay fraction. The generalization of the model to new set of data other than the training set of data is discussed which is required for successful application of a model. A detailed study of the relative performance of the computational intelligence techniques has been carried out based on different statistical performance criteria.     

Comparative study of surrogate models for groundwater contamination source identification at DNAPL-contaminated sites

Knowledge of groundwater contamination sources is critical for effectively protecting groundwater resources, estimating risks, mitigating disaster, and designing remediation strategies. Many methods for groundwater contamination source identification (GCSI) have been developed in recent years, including the simulation–optimization technique. This study proposes utilizing a support vector regression (SVR) model and a kernel extreme learning machine (KELM) model to enrich the content of the surrogate model. The surrogate model was itself key in replacing the simulation model, reducing the huge computational burden of iterations in the simulation–optimization technique to solve GCSI problems, especially in GCSI problems of aquifers contaminated by dense nonaqueous phase liquids (DNAPLs). A comparative study between the Kriging, SVR, and KELM models is reported. Additionally, there is analysis of the influence of parameter optimization and the structure of the training sample dataset on the approximation accuracy of the surrogate model. It was found that the KELM model was the most accurate surrogate model, and its performance was significantly improved after parameter optimization. The approximation accuracy of the surrogate model to the simulation model did not always improve with increasing numbers of training samples. Using the appropriate number of training samples was critical for improving the performance of the surrogate model and avoiding unnecessary computational workload. It was concluded that the KELM model developed in this work could reasonably predict system responses in given operation conditions. Replacing the simulation model with a KELM model considerably reduced the computational burden of the simulation–optimization process and also maintained high computation accuracy.     

基于Ricker子波核的支持向量回归方法及其在地震勘探记录去噪处理中的应用

针对地震勘探中强随机噪声的去噪问题,引进支持向量回归方法,提出并证明一种新的Ricker子波核函数。支持向量回归采用核映射的基本思想,基于结构风险最小化原则,将回归问题转化为一个二次规划问题。对单道记录或多道记录中任选道的仿真实验表明,与传统的基于径向基核函数的支持向量回归及褶积滤波方法相比,使用本方法去噪后的同相轴更为清晰,波形恢复得更好,信噪比也较高,因此有可能将其应用于地震勘探记录的去噪处理中。     

Landslide susceptibility mapping using GIS-based multi-criteria decision analysis,support vector machines,and logistic regression

Identification of landslides and production of landslide susceptibility maps are crucial steps that can help planners, local administrations, and decision makers in disaster planning. Accuracy of the landslide susceptibility maps is important for reducing the losses of life and property. Models used for landslide susceptibility mapping require a combination of various factors describing features of the terrain and meteorological conditions. Many algorithms have been developed and applied in the literature to increase the accuracy of landslide susceptibility maps. In recent years, geographic information system-based multi-criteria decision analyses (MCDA) and support vector regression (SVR) have been successfully applied in the production of landslide susceptibility maps. In this study, the MCDA and SVR methods were employed to assess the shallow landslide susceptibility of Trabzon province (NE Turkey) using lithology, slope, land cover, aspect, topographic wetness index, drainage density, slope length, elevation, and distance to road as input data. Performances of the methods were compared with that of widely used logistic regression model using ROC and success rate curves. Results showed that the MCDA and SVR outperformed the conventional logistic regression method in the mapping of shallow landslides. Therefore, multi-criteria decision method and support vector regression were employed to determine potential landslide zones in the study area.     

加权支持向量回归机及其在水质预测中的应用

支持向量机是一种基于结构风险最小化原理的学习技术，也是一种新的具有很好泛化性能的回归方法。本文对用于回归估计的标准支持向量机加以改进，提出了一种新的用于回归估计的支持向量机学习算法，针对各样本重要性的差异，给各个样本的惩罚系数和误差要求赋予不同权重，并利用加权支持向量回归机的理论及其算法构建水质预测模型。实验结果表明，该方法对水质具有较好的预测效果。     

基于时间序列与人工蜂群支持向量机的滑坡位移预测研究

总结以往滑坡预测方法存在的诸多不足,针对滑坡监测位移-时间曲线特点,本文提出了一种基于时间序列的人工蜂群算法（ABC）与支持向量回归机（SVR）相结合的滑坡位移预测方法。以三峡库区白水河滑坡为例,通过对滑坡位移、降雨、库水位等因素的分析,研究影响滑坡位移变化的因素。用时间序列加法模型和移动平均法将滑坡位移分解为趋势项和周期项。以多项式最小二乘法拟合滑坡位移趋势项,用人工蜂群支持向量机模型对滑坡位移周期项进行训练和预测。通过灰色系统关联分析法计算多项因子与滑坡位移周期项之间的关联性。最终的滑坡总位移预测值为周期项预测值与趋势项预测值之和。与BP神经网络、PSO-SVR模型方法相比,该方法在滑坡位移预测中有更高的精度,在防灾减灾工作中有较好的推广应用前景。     

Prediction of porosity and water saturation using pre-stack seismic attributes: a comparison of Bayesian inversion and computational intelligence methods

Rock physical parameters such as porosity and water saturation play an important role in the mechanical behavior of hydrocarbon reservoir rocks. A valid and reliable prediction of these parameters from seismic data is essential for reservoir characterization, management, and also geomechanical modeling. In this paper, the application of conventional methods such as Bayesian inversion and computational intelligence methods, namely support vector regression (SVR) optimized by particle swarm optimization (PSO) and adaptive network-based fuzzy inference system-subtractive clustering method (ANFIS-SCM), is demonstrated to predict porosity and water saturation. The prediction abilities offered by Bayesian inversion, SVR-PSO, and ANFIS-SCM were presented using a synthetic dataset and field data available from a gas carbonate reservoir in Iran. In these models, seismic pre-stack data and attributes were utilized as the input parameters, while the porosity and water saturation were the output parameters. Various statistical performance indexes were utilized to compare the performance of those estimation models. The results achieved indicate that the ANFIS-SCM model has strong potential for indirect estimation of porosity and water saturation with high degree of accuracy and robustness from seismic data and attributes in both synthetic and real cases of this study.     

Prediction of landslide displacement with step-like behavior based on multialgorithm optimization and a support vector regression model

Landslide prediction is important for mitigating geohazards but is very challenging. In landslide evolution, displacement depends on the local geological conditions and variations in the controlling factors. Such factors have led to the “step-like” deformation of landslides in the Three Gorges Reservoir area of China. Based on displacement monitoring data and the deformation characteristics of the Baishuihe Landslide, an additive time series model was established for landslide displacement prediction. In the model, cumulative displacement was divided into three parts: trend, periodic, and random terms. These terms reflect internal factors (geological environmental, gravity, etc.), external factors (rainfall, reservoir water level, etc.), and random factors (uncertainties). After statistically analyzing the displacement data, a cubic polynomial model was proposed to predict the trend term of displacement. Then, multiple algorithms were used to determine the optimal support vector regression (SVR) model and train and predict the periodic term. The results showed that the landslide displacement values predicted based on data time series and the genetic algorithm (GA-SVR) model are better than those based on grid search (GS-SVR) and particle swarm optimization (PSO-SVR) models. Finally, the random term was accurately predicted by GA-SVR. Therefore, the coupled model based on temporal data series and GA-SVR can be used to predict landslide displacement. Additionally, the GA-SVR model has broad application potential in the prediction of landslide displacement with “step-like” behavior.     

Spatial prediction of landslide susceptibility in western Serbia using hybrid support vector regression (SVR) with GWO,BAT and COA algorithms

In this study, we developed multiple hybrid machine-learning models to address parameter optimization limitations and enhance the spatial prediction of landslide susceptibility models. We created a geographic information system database, and our analysis results were used to prepare a landslide inventory map containing 359 landslide events identified from Google Earth, aerial photographs, and other validated sources. A support vector regression (SVR) machine-learning model was used to divide the landslide inventory into training (70%) and testing (30%) datasets. The landslide susceptibility map was produced using 14 causative factors. We applied the established gray wolf optimization (GWO) algorithm, bat algorithm (BA), and cuckoo optimization algorithm (COA) to fine-tune the parameters of the SVR model to improve its predictive accuracy. The resultant hybrid models, SVR-GWO, SVR-BA, and SVR-COA, were validated in terms of the area under curve (AUC) and root mean square error (RMSE). The AUC values for the SVR-GWO (0.733), SVR-BA (0.724), and SVR-COA (0.738) models indicate their good prediction rates for landslide susceptibility modeling. SVR-COA had the greatest accuracy, with an RMSE of 0.21687, and SVR-BA had the least accuracy, with an RMSE of 0.23046. The three optimized hybrid models outperformed the SVR model (AUC = 0.704, RMSE = 0.26689), confirming the ability of metaheuristic algorithms to improve model performance.     

A comparison of two models to predict soil moisture from remote sensing data of RADARSAT II

This study investigates the performance of empirical and semiempirical models to predict soil moisture from the data of RADARSAT II synthetic aperture radar (SAR) for the Yarmouk basin in Jordan. Data of SAR were obtained for May and June 2010 and were processed to obtain backscatter (σ ^o ) data for the study area. Results showed significant correlations between soil moisture content (m _v ) and horizontally polarized σ ^o , with coefficient of determination (R ²) of 0.64. The root mean square error for the SAR volumetric soil moisture content was 0.09 and 0.06 m³/m³ for the empirical and semiempirical regression models, respectively. Both models had different clustering patterns in the soil moisture maps in the study area. The spatial agreement between maps of soil moisture was in the range of 55 to 65 % when the maps were reclassified based on intervals of 5 % m _v for both models. In terms of soil moisture interval, both models showed that most of soil moisture changes between the two images (dates) were in the range of ±5 %. Some high differences in ?m _v were observed between the two models. These were mainly attributed to the non-inverted pixels in the soil moisture maps produced by the semiempirical model. Therefore, this model may be applied for a limited range of soil moisture prediction. The use of regression model could predict a wider range for soil moisture when compared with the semiempirical model. However, more work might be needed to improve the empirical model before scaling it up to the whole study area.     

Improved RMR Rock Mass Classification Using Artificial Intelligence Algorithms

Rock mass classification systems such as rock mass rating (RMR) are very reliable means to provide information about the quality of rocks surrounding a structure as well as to propose suitable support systems for unstable regions. Many correlations have been proposed to relate measured quantities such as wave velocity to rock mass classification systems to limit the associated time and cost of conducting the sampling and mechanical tests conventionally used to calculate RMR values. However, these empirical correlations have been found to be unreliable, as they usually overestimate or underestimate the RMR value. The aim of this paper is to compare the results of RMR classification obtained from the use of empirical correlations versus machine-learning methodologies based on artificial intelligence algorithms. The proposed methods were verified based on two case studies located in northern Iran. Relevance vector regression (RVR) and support vector regression (SVR), as two robust machine-learning methodologies, were used to predict the RMR for tunnel host rocks. RMR values already obtained by sampling and site investigation at one tunnel were taken into account as the output of the artificial networks during training and testing phases. The results reveal that use of empirical correlations overestimates the predicted RMR values. RVR and SVR, however, showed more reliable results, and are therefore suggested for use in RMR classification for design purposes of rock structures.