基于极限学习机的GF-2影像分类

遥感图像分类是提取图像有效信息过程中重要的一部分,为了探寻最优的分类方法,许多机器学习算法逐步应用于遥感分类中。极限学习机（extreme learning machine,ELM）以其高效、快速和良好的泛化性能在模式识别领域得到广泛应用。本文采用训练速度快、运算量小的极限学习机算法与支持向量机（support vector machines,SVM）算法和最大似然法进行分类对比,对高分辨率遥感图像进行分类,分析极限学习机算法对于遥感图像分类的准确度等性能。选取吉林省长春市部分区域的GF-2遥感数据,将融合后的影像设置为原始数据,利用3种方法进行分类。研究结果表明,极限学习机算法分类图像总体分类精度达到85%以上,kappa系数达到0.718,与其他分类方法相比分类准确度较高,且极限学习机运行时间比支持向量机运行时间约短2 480 s,约为支持向量机运行时间的1/8,因此具有良好的性能和实用价值。     

Predictive Modeling of Mining Induced Ground Subsidence with Survival Analysis and Online Sequential Extreme Learning Machine

Mining induced land subsidence is one of the most hazardous geological phenomenon. Predictive modeling of the ground subsidence has attracted increased interest and is crucial to the hazard prevention. In this research, a data-driven approach integrated with survival analysis to model the mining-induced subsidence is studied. The data used in this research is collected in Fuxin, Liaoning Province, China and it contains multiple variables from different subsided locations. First, a survival analysis is conducted using the Cox proportional hazard model to evaluate the importance of variables considered. p values of all variables are computed and the important variables are selected. Next, data-driven models including k-nearest neighbor, support vector machine, back-propagated neural network, random forest, extreme learning machine, and online sequential extreme learning machine are constructed to predict the subsidence values and horizontal movement. Two evaluation matrices namely MAPE and RMSE are introduced to evaluate the performances of the data-driven models. Computational results demonstrate that online sequential extreme learning machine is capable of accurately predict the mining induced subsidence and surface deformation.     

On the Investigations of Resilient Modulus of Residual Tropical Gravel Lateritic Soils from Senegal (West Africa)

Investigations on the mechanical behavior of compacted gravel lateritic soils have been the subject of several studies. Used as road materials, soils tests were mainly performed using standard tests. Static loads as unconfined compression test (UCT) remain the only engineering approach used. Alternative testing techniques can be chosen as supplementary tests for characterizing pavement materials. These researches were conducted so as to determine the response of these particular and problematic soils in its compacted form with road traffic loads. This paper presents the results of research conducted to investigate the effect of the soil compacity on the resilient modulus of lateritic soils. The influence of the percentage of cement added so as to stabilize each sample at the optimum modified proctor (OPM) State was also determined. Soil big specimens of around 180 mm diameter (with length to diameter aspect ratio of 2:1) were prepared according to the standard procedure described by AASHTO T 307 and then were subjected to repeated load triaxial tests. The models used, analyzed and developed in this paper are mainly the Andrei and the Uzan–Witczak universal model. Test results showed that the specimen compacity has no significant influence on the resilient modulus of the investigated gravel lateritic soils. Soil specimens with variation of the percentage of cement added exhibited the highest resilient modulus values while the specimens with variation of the compacity exhibited the lowest values. The resilient modulus variation seems to be independent of the level of stress.     

基于极限学习机的遥感地球化学反演模型

地球化学勘查研究涉及大量采样工作,但在工作环境恶劣人们难以到达的地区,大范围、大比例尺的地球化学数据极难获取。本文基于极限学习机（ELM）构建遥感地球化学反演模型,弥补因为区域数据不足导致的找矿工作困难。依据偏最小二乘回归（PLSR）方法选取与地球化学数据相关性强的遥感影像成分,并根据极限学习机建立地球化学数据与遥感影像之间的非线性对应关系来获取未知地球化学异常,以此来指导找矿工作。实验中,选取研究区铜元素1：20万土壤地球化学数据与Landsat 8 OLI遥感影像进行反演分析。实验结果表明,基于ELM的遥感地球化学反演所取得的异常分布与已知矿点具有很好的对应度,验证了本文所提出模型的有效性。     

集成极限学习机的中小河流洪水预报方法研究

为利用水文现象相似性和极限学习机(ELM)集成学习提高洪水预报精度,提出了一种基于相似度匹配的集成ELM洪水预报方法(SM-ELM)。方法首先从多个ELM模型中,为每一个训练样本找到最优的ELM模型,然后从训练集中,为测试样本匹配出最相似的前k个训练样本,最后利用这k个训练样本分别对应的最优ELM模型,对测试样本采用加权平均法进行集成预报。为证明提出方法的可行性和有效性,以昌化流域的历史洪水为例进行了验证。结果表明,相对于单个ELM,集成ELM模型能有效地提高预测精度。从均方根误差上看,集成ELM模型性能比单个ELM模型提升了10%~15%。在三种集成方法中,SM-ELM能够以较少的模型数量获得较高且稳定的预报精度。     

复合地基复合模量的分析和应用

讨论了多桩型复合地基复合模量的基本概念，通过对具体工程的沉降计算和实测结果对比分析，对复合模量的正确分析和应用提出了一些建议。     

The casualty prediction of earthquake disaster based on Extreme Learning Machine method

Natural Hazards - In the prediction of casualties of earthquake disaster, the traditional prediction method requires strict sample data, and it is necessary to manually set a large number of...     

Effect of Matric Suction on Resilient Modulus of Compacted Aggregate Base Courses

This research was conducted to investigate the effect of matric suction on resilient modulus of unbound aggregate base courses. The study characterized the water characteristic curves and resilient modulus versus matric suction relationships of aggregate base courses that were compacted at different water contents and between 98 and 103 % of the modified Proctor density. The soil–water characteristic curve (SWCC) and the relationship between resilient modulus (M _r ) and matric suction (ψ) were established for different unbound granular and recycled asphalt pavement materials. This relationship is important for predicting changes in modulus due to changes in moisture of unbound pavement materials. Resilient modulus tests were conducted according to the National Cooperative Highway Research Program (NCHRP) 1-28A procedure at varying water contents, and the measured SWCC was used to determine the corresponding matric suction. Three reference summary resilient moduli (SRM) were considered: at optimum water content, optimum water content +2 % and optimum water content ?2 %. The Bandia and Bargny limestones are characterized by a higher water-holding capacity which explains why the modulus of limestone was more sensitive to water content than for basalt or quartzite. Limestones tend to be more sensitive to changes in water content and thus to matric suction. The shape of the SWCC depends on the particle size distribution and the cementation properties from dehydration of the aggregates. Material properties required as input to the Mechanistic-Empirical Pavement Design Guide (M-EPDG) to predict changes in resilient modulus in response to changes in moisture contents in the field were determined for implementation in the M-EPDG process. Results show that the SRM was more correlated with matric suction than with compaction water content (for resilient modulus testing). The empirical models commonly used to predict the SWCC such as the Perera et al. (Prediction of the SWCC based on grain-size-distribution and index properties. GSP 130 Advances in Pavement Engineering, ASCE, 2005) and the M-EPDG (NCHRP in Guide for mechanistic-empirical design of pavement structures. National cooperative highway research program. ARA, Inc., ERES Consultants Division, Champaign, IL, 2004) models tend to underestimate the SWCC and cannot provide reasonable estimation. SRM normalized with respect to the SRM at the optimum water content varied linearly with the logarithm of matric suction. Empirical relationships between SRM and matric suction on semi-logarithmic scale were established and are reported.     

Shear Modulus and Damping Ratio of Organic Soils

The paper presents results from a laboratory investigation into the dynamic properties of natural intact and model organic soils by means of resonant-column tests. The natural intact organic soils were sands, cohesive soils and peats with varying content of calcium carbonate. The model organic soils were formed in laboratory by mixing kaolinite and paper pulp. The influence of various soil parameters, such as strain level, confining stress, void ratio, plasticity index, organic content and secondary consolidation time on shear modulus, G, and damping ratio, DT, is presented and discussed. The test results on natural organic soils show that only high organic contents (OC ≥ 25%) have significant influence on G and DT at both small and high shear strains. For the model organic soils, however, a significant influence of even lower values of organic content (5% ≤ OC ≤ 20%) on G at small strains and DT at both small and high strains is observed.     

Machine Learning Based Predictive Modeling of Debris Flow Probability Following Wildfire in the Intermountain Western United States

It has been recognized that wildfire, followed by large precipitation events, triggers both flooding and debris flows in mountainous regions. The ability to predict and mitigate these hazards is crucial in protecting public safety and infrastructure. A need for advanced modeling techniques was highlighted by re-evaluating existing prediction models from the literature. Data from 15 individual burn basins in the intermountain western United States, which contained 388 instances and 26 variables, were obtained from the United States Geological Survey (USGS). After randomly selecting a subset of the data to serve as a validation set, advanced predictive modeling techniques, using machine learning, were implemented using the remaining training data. Tenfold cross-validation was applied to the training data to ensure nearly unbiased error estimation and also to avoid model over-fitting. Linear, nonlinear, and rule-based predictive models including naïve Bayes, mixture discriminant analysis, classification trees, and logistic regression models were developed and tested on the validation dataset. Results for the new non-linear approaches were nearly twice as successful as those for the linear models, previously published in debris flow prediction literature. The new prediction models advance the current state-of-the-art of debris flow prediction and improve the ability to accurately predict debris flow events in wildfire-prone intermountain western United States.     

基于机器学习的水位流量关系模型参数估计

为了克服经典水位流量关系模型在刻画河流动态变化特性时所存在的局限性,提出采用局部加权回归算法估计模型参数;为了提高参数估计精度以及流量的计算效率,提出一种聚类树加权回归方法.首先对训练样本进行聚类,然后使用k-最近邻方法将新的水位样本划分进最恰当的聚类中,最后估计河流日流量.该方法在估计过程中,避免了不相关信息的干扰,从而提高了日流量数据估计的效率和精度.利用某水文站的实测数据对方法进行测试,仿真结果表明,方法估计精度高,为水位流量关系模型参数估计提供了新的有效方法.     

A Three-Dimensional Geological Structure Modeling Framework and Its Application in Machine Learning

Mathematical Geosciences - Three-dimensional geological structure analysis is fundamental to geoscientific research. With the application of artificial intelligence in geological structure...     

冻融循环作用下路基土回弹模量试验研究

为研究季节冻土地区冻融循环作用对路基土回弹模量的影响,以2种路基土作为研究对象,通过风干和加湿的方法,使高50mm、直径为50mm无侧限圆柱体试件达到5种预期含水率;采用承载板法进行回弹模量试验,分析了路基土在0～12次冻融循环作用下,回弹模量随冻融循环次数变化的规律.结果表明:随着冻融循环次数的增加,土体的回弹模量减小,到第6次冻融循环后土体回弹模量衰减基本稳定.在进行季节冻土地区路面设计时,应考虑冻融循环作用对路基土的影响,建议选取第6次冻融循环后的土体回弹模量作为路基强度设计值.     

Prediction of Resilient Modulus from Post-compaction Moisture Content and Physical Properties Using Support Vector Regression

The present study assesses the use of support vector machine regression to predict the variation of resilient modulus with post-compaction moisture content of soils commonly encountered in Oklahoma, Pennsylvania and Wisconsin. Results show the prediction model using the support vector regression (SVR) approach is a function of degree of saturation, moisture content and plasticity index. The developed model is compared to current models in the literature. Results indicate the proposed SVR model gives more accurate values than current regression models. This model will better predict changes in the bearing capacity of pavements due to seasonal variations of moisture content.     

不均匀地基对北大红楼变形破坏的影响分析

依据对北大红楼的工程地质调查,其地基条件为不均匀地基土,借助FLAC^2D有限差分程序分析不均匀地基对北大红楼变形破坏的影响,发现北大红楼上部建筑出现的宏观变形破坏特征与下部地基岩土层性质、厚度、空间分布状况等密切相关。分析结果对于北大红楼维修加固设计具有重要的指导意义。     

Semi-empirical Model for Estimating the Small-Strain Shear Modulus of Unsaturated Non-plastic Sandy Soils

The shear modulus at small-strain, G _max is a maximum value of shear modulus for a given stress state and void ratio, and is a key parameter to evaluate the dynamic response of geotechnical structures. However, the laboratory testing procedures for determining G _max are time-consuming, cumbersome and require elaborate equipment especially for unsaturated soil samples. A semi-empirical model is proposed in this paper that can be used to estimate the variation of G _max with respect to matric suction for non-plastic sandy soils (i.e. I _p  = 0 %). The proposed model uses the Soil–Water Characteristic Curve (SWCC) and the shear modulus at saturation condition along with two fitting parameters ζ and ξ. The proposed model permits estimation of the variation of G _max with respect to matric suction over different zones of the SWCC (i.e. boundary effect, transition, and residual zones) for various non-plastic sandy soils. The fitting parameters ζ and ξ required for the proposed semi-empirical model can be estimated from simple relationships derived from the grain size distribution curve.     

路基土动荷载下力学行为研究进展

从试验研究、模型和本构关系研究两大部分入手,介绍了路基土在动荷载作用下的塑性及回弹行为、安定行为与临界应力水准及其判别、应变速率、固结沉降与固结度、动力本构模型、孔压模型、应变软化、应交硬化等基本特征,对比分析了国内外对路基土动荷载下力学行为研究的最新进展,总结了当前路基土研究中存在的主要问题,并提出了今后研究的发展趋势.     

基于机器学习的矿物智能识别方法研究进展与展望

矿物智能识别是地球科学与信息科学的重要交叉方向,显示出强大的生命力.本文在调研国内外研究动态的基础上,把矿物智能识别划分为4个阶段,即矿物采集、数据获取、模型构建、分类判别;根据测试方法和获得的数据类型,把矿物智能识别分为基于化学成分、显微光学图片、光谱分析的3条基本路线;总结了应用于矿物智能识别的机器学习方法和技术,...