空间聚类的研究现状及其应用

作为空间数据挖掘的一种重要手段，空间聚类目前已在许多领域得到了应用。文章在对已有空间聚类分析方法概括和总结的基础上，结合国家卫星气象中心高分辨率有限区域分析预报系统产品中的数值格点预报(HLAFS)值，运用K—均值法对影响青藏高原上中尺度对流系统(MCS)移动的散度场进行了研究，得到了一些有意义的结论。     

PNN及其在储层物性参数预测中的应用

概率神经网络(PNN)实际上是一种数学插值手段，它的实现引用了神经网络的结构但它通过研究数学规则，可以比多层前馈神经网络更好地实现插值功能，因而具有潜在的优势，这里简要介绍了PNN的基本原理并用于储层物性参数的预测。实践表明该方法是有效的。     

基于AlexNet网络的二维找矿预测——以松桃—花垣地区沉积型锰矿为例

在大数据的时代背景下,地质大数据逐渐趋于复杂化的模式与其间的空间关联性为基于机器学习算法的矿产资源定量预测带来了更大的挑战。利用深度卷积网络算法优异的分析性能来提取不同成矿条件下多种二维要素图层的空间分布特征与关联性是一项非常有意义的探索性实验。以松桃—花垣地区沉积型锰矿为例,利用深度卷积神经网络模型AlexNet挖掘Mn元素、沉积相、大塘坡组出露、断裂及水系的空间分布与锰矿矿床的就位空间的耦合相关性,以及不同的控矿要素之间的相关性,以此训练出二维矿产预测分类模型。经过训练后,可以得到验证准确率88.89%,召回率为66.67%,损失值0.08的深度卷积神经网络分类模型。应用该模型对未知区进行二维成矿预测,共圈定出91、96、154、184号4个成矿远景区,其中91号和154号的区域含矿概率为1,96号含矿概率为0.5。由此可见,预测区具有很大概率存在尚未发现的矿床。     

基于Alpha稳定分布概率神经网络的围岩稳定性分类研究EI北大核心CSCD

工程围岩是一种高度非线性的复杂动态系统,其影响因素众多,单一的评价指标已不能准确描述围岩分类情况。目前,综合考虑多种指标评价围岩分类的方法很多,但围岩评价指标之间或多或少存在一定的相关性,其评价指标中存在一些服从非高斯分布的指标,无法满足概率神经网络(PNN)样本层中采用高斯分布作径向基函数的要求,因此,提出一种对称Alpha稳定分布(SaS)。SaS有更广泛的数学表达,其径向对称特性还可充当PNN样本层中高斯分布。在SaS的基础上,建立广州抽水蓄能电站二期工程围岩分类评价的SaS-PNN模型。预测结果表明,SaS-PNN模型具有良好的预测效果,其误判率为为4.55%。可为地下工程围岩分类评价提供一种新思路。     

采空区移动变形的径向基函数神经网络概率积分法(RBF)反演

借助部分矿区地表移动实测数据,选用径向基函数(RBF)神经网络对概率积分法的计算参数进行反演,采用K层交叉验证对模型精度进行优化,并与BP神经网络模型和SVM模型预测结果进行对比,发现RBF模型精度均优于BP模型和SVM模型,且其稳定性较好。可为概率积分法预测评估采空区地表移动变形范围提供一种可靠的方法。     

基于目标检测的地质异常信息提取——以湖南香花岭地区为例

卷积神经网络在图像识别领域处于领先地位,在目标检测方面的应用也越来越广泛,其特点是可以提取点与点之间的相关关系。二维地质图中,点与点之间往往存在特定的空间关系,将卷积神经网络技术应用于地质异常信息的提取有其重要性。在讨论卷积神经网络以及基于该技术目标检测算法(YOLO)的基础上,以湖南香花岭地区为例,提取与锡矿成矿相关的构造信息并进行分析,结果该方法能够覆盖原有矿点,有效地定义点与点之间的相关关系,描述点与点之间的空间相关性,可靠地提取与成矿有关的构造线密度信息,在花岗岩体与构造复杂地区圈出地质异常信息。     

融入先验知识的贝叶斯神经网络分类器

提高神经网络模型推广性的关键是控制模型的复杂度，给出了一种基于贝叶斯推理的神经网络方法，它能自然地融入关于模型的先验知识，与观察到的数据样本相综合来控制神经网络模型中不同部分的复杂度。使用马尔可夫链蒙特卡罗模拟可获得模型参数的后验分布，预测分类是许多个以各自后验分布为权的马尔可夫链上的模型样本的平均。在二个实际分类问题应用中与常规神经网络方法进行了对比分析。     

不同积云对流参数化方案对黑河流域降水模拟的影响

使用NCEP-fnl再分析资料作为黑河流域高分辨率区域气候模式的初始场和边界场,利用该模式中常用的3种积云对流参数化方案:Grell,Bett-Miller(BM)和不采用积云对流参数化方案(NON)对黑河流域进行2000年1月1日至12月31日的积分试验,重点考察水平分辨率在3 km条件下不同积云对流参数化方案对黑河流域降水模拟的敏感性。结果表明:①卫星遥感反演的黑河流域的降水较观测台站降水偏少,卫星遥感反演日降水与观测台站日降水的相关系数达到0.34,相关系数通过99%置信度检验;②模式采用3种参数化方案都能够较好地模拟出年降水空间分布以及不同区域日平均降水随时间演变,与观测之间的相关系数都通过99%置信度检验;③对于黑河流域来说,在水平分辨率为3 km条件下区域气候模式采用Grell积云对流参数化方案较其他2种方案无论从空间和时间演变来说均更加接近观测。     

概率神经网络储层流体密度反演及应用

概率神经网络是一种基于概率统计思想的神经网络,利用概率神经网络进行储层的流体密度反演,通过它的非线性扩展进行多个属性的优选组合,完成神经网络的训练学习和概率估算,有效地剔除个别数据的不利影响,使反演过程更加稳定,减少反演结果的多解性.川西某气藏的概率神经网络储层流体密度反演结果表明,该反演方法准确性很高,反演结果与实际...     

考虑参数空间变异性多层土坡系统可靠度分析

多层土坡在岩土工程实际中十分常见,不仅土体参数存在一定的空间变异性,而且土体框架呈现明显的层状分布特征,然而目前对考虑土体参数空间变异性的多层土坡稳定可靠度研究的远远不够。提出了基于多重响应面边坡系统可靠度分析的蒙特卡洛模拟（MCS）方法,给出了计算流程图,系统地研究了考虑土体参数空间变异性的多层土坡系统可靠度问题。结果表明,提出方法能够有效地分析考虑参数空间变异性低失效概率水平的多层土坡系统可靠度问题,并且具有较高的参数敏感性分析计算效率。     

Sequence stratigraphy, structural style, and age of deformation of the Malaita accretionary prism (Solomon arc–Ontong Java Plateau convergent zone)

Possibilities for the fate of oceanic plateaus at subduction zones range from complete subduction of the plateau beneath the arc to complete plateau–arc accretion and resulting collisional orogenesis. Deep penetration, multi-channel seismic reflection (MCS) data from the northern flank of the Solomon Islands reveal the sequence stratigraphy, structural style, and age of deformation of an accretionary prism formed during late Neogene (5–0 Ma) convergence between the 33-km-thick crust of the Ontong Java oceanic plateau and the 15-km-thick Solomon island arc. Correlation of MCS data with the satellite-derived, free-air gravity field defines the tectonic boundaries and internal structure of the 800-km-long, 140-km-wide accretionary prism. We name this prism the “Malaita accretionary prism” or “MAP” after Malaita, the largest and best-studied island exposure of the accretionary prism in the Solomon Islands. MCS data, gravity data, and stratigraphic correlations to islands and ODP sites on the Ontong Java Plateau (OJP) reveal that the offshore MAP is composed of folded and thrust faulted sedimentary rocks and upper crystalline crust offscraped from the Solomon the subducting Ontong Java Plateau (Pacific plate) and transferred to the Solomon arc. With the exception of an upper, sequence of Quaternary? island-derived terrigenous sediments, the deformed stratigraphy of the MAP is identical to that of the incoming Ontong Java Plateau in the North Solomon trench.We divide the MAP into four distinct, folded and thrust fault-bounded structural domains interpreted to have formed by diachronous, southeast-to-northwest, and highly oblique entry of the Ontong Java Plateau into a former trench now marked by the Kia–Kaipito–Korigole (KKK) left-lateral strike-slip fault zone along the suture between the Solomon arc and the MAP. The structural style within each of the four structural domains consists of a parallel series of three to four fault propagation folds formed by the seaward propagation of thrust faults roughly parallel to sub-horizontal layering in the upper crystalline part of the OJP. Thrust fault offsets, spacing between thrusts, and the amplitude of related fault propagation folds progressively decrease to the west in the youngest zone of active MAP accretion (Choiseul structural domain). Surficial faulting and folding in the most recently deformed, northwestern domain show active accretion of greater than 1 km of sedimentary rock and 6 km, or about 20%, of the upper crystalline part of the OJP. The eastern MAP (Malaita and Ulawa domains) underwent an earlier, similar style of partial plateau accretion. A pre-late Pliocene age of accretion (3.4 Ma) is constrained by an onshore and offshore major angular unconformity separating Pliocene reefal limestone and conglomerate from folded and faulted pelagic limestone of Cretaceous to Miocene age. The lower 80% of the Ontong Java Plateau crust beneath the MAP thrust decollement appears unfaulted and unfolded and is continuous with a southwestward-dipping subducted slab of presumably denser plateau material beneath most of the MAP, and is traceable to depths >200 km in the mantle beneath the Solomon Islands.     

基于BP神经网络的薄互层储层预测

根据地震属性来进行储层预测的研究,对于寻找油气具有十分重要的意义。而由于地下地质情况千变万化,不确定的因素太多,对这种多自变量与因变量的复杂关系模拟问题,神经网络技术是目前较成熟、实际应用效果也较好的方法之一。用BP神经网络来预测薄互层储层厚度,它可以建立属性参数与预测目标之间的高度非线性映射,并应用于具体的实例,为薄互层储层预测提供了新的思路。     

河北坝上高原如意河流域风积沙厚度空间展布预测方法

不同的空间插值方法会对风积沙厚度空间分布预测精度产生重要影响。基于河北承德坝上高原东部如意河流域中游152组风积沙厚度数据,应用径向基函数人工神经网络(RBF-ANN)插值方法探究如意河流域中游风积沙厚度空间分布特征,并与地统计插值方法中的不同模型函数、确定性插值中的不同插值方法进行预测误差和计算结果对比分析。结果表明,在地统计插值方法中,经典贝叶斯克里金插值-幂半变异函数(EBK- Power)的插值效果最佳;在确定性插值方法中,径向基函数(RBF)插值效果最佳;RBF-ANN插值较EBK- Power和RBF方法在平均绝对误差(MAE)上的改进大于30%,在均方根误差(RMSE)上的改进大于20%,RBF-ANN插值更适用于如意河流域风积沙厚度空间分布预测。     

Study on Environmental and Hydrological Effects of Thermokarst Lakes in Permafrost Regions of the Qinghai-Tibet Plateau

Thermokarst lake is the most visible morphologic landscape developing during the process of permafrost degradation, and it is still an international hot topic in permafrost research. The climate warming, and the consequent degradation of the permafrost on the Qinghai-Tibet Plateau aggravate thermokarst lake development. The permafrost is normally considered as an aquiclude, and the permafrost degradation, especially when the permafrost is completely thawed by a thermokarst lake, might influence regional ground water. Therefore, a research program focusing on environmental and hydrological effects of thermokarst lakes in permafrost regions of the Qinghai-Tibet Plateau was started and supported by the National Natural Science Foundation of China. The work proposed by the application includes: To analysis the spatial and temporal distribution rule of thermokarst lakes in the Qinghai-Tibet Engineering Corridor (QTEC) under the climate change and engineering activities, and to evaluate the ecological environment effects through remote sensing and field investigation; to reveal the main factors influencing a typical thermokarst lake and its hydrothermal condition, and to elucidate the conversion relationship between the thermokarst lake and the groundwater with hydrological and isotope tracer tests; to make an analysis of the influences of different lake stage and size on regional permafrost, hydrological conditions and ecological environment through numerical simulation and statistical modelling, considering the relationships between the thermokarst lake and the ground water level. The research results will help to accurately assess regional permafrost ecological environment evolution and trend prediction, and to reasonably understand the impact factors of the permafrost hydrological evolution and its response mechanism to the ecological environment in the river source regions of the Qinghai-Tibet Plateau. In this paper, the research status analysis, the main research contents, research objectives and prospects were introduced so as to provide some references for related researchers and engineers.     

Using monitoring data of surface soil to predict whole crop-root zone soil water content with PSO-LSSVM,GRNN and WNN

Drought is a significant disaster in Beijing and it is important to find a method to assess the drought condition. First, this paper collected data of 85 soil monitoring stations in Beijing, such as soil dry bulk densities, saturated water contents, field capacities. Then, spatial variability characteristics of soil physics parameters were investigated by GIS and other three factors, 10 cm soil moisture content, organic matter and saturated water content which notably influenced soil moisture were extracted by Principal Component Analysis (PCA). Furthermore, four different nonlinear methods were put forward to predict crop-root zone soil water. 15555 single daily data from 2011 were used in parameters determination, while 15470 double daily data were used to test. The result showed that the Least Square Support Vector Machine coupling Particle Swarm Optimization Algorithm (PSO-LSSVM) (R ²?=?0. 875) did better than BP Neural Network (R ²?=?0. 840), Generalized Regression Neural Network (GRNN) (R ²?=?0. 850) and Wavelet Neural Network (WNN) (R ²?=?0. 853). As so the POS-LSSVM method was used to evaluate the drought conditions from October 2010 to March 2011 of Beijing, and the result showed that from October 2010 to January 2011, the drought conditions were getting increasingly worse while later relieved from January 2011 to March 2011.     

Research Status and Prospect of the Interaction Between Tibetan Plateau Shear Line and Tibetan Plateau Vortex

The Tibetan Plateau Shear Line (TPSL) is usually accompanied by the Tibetan Plateau Vortex (TPV) and this phenomenon is one of the assembled weather systems over the Tibetan Plateau (TP) and its surrounding areas. This assembled system plays a very important role in the high impact weather process in the TP and East China. We reviewed the research history and progress of TPVs and TPSLs, and mainly discussed the relationship and interaction mechanism of them. According to the latest research achievement of TPSLs and its relationship with TPVs, the development and application trends of related theory and methods, we proposed several notable new research directions in the field of this study. It is not clear for the relationship and the physical mechanism of the interaction between TPSLs and TPVs as well as some high impact weather initiated by them currently. Therefore, this research work is really quite important for theoretical development of weather dynamics of the TP, and is expected to provide a theoretical guide for severe weather analysis and forecast over the TP and its neighborhood.     

Conditional LAS stochastic simulation of regionalized variables in random fields

Ore reserves forecasts are required to aid in investment decisions, mine design and valuation, short and long term production plans and proper and efficient mill design. In random multivariable fields with limited data and high levels of uncertainty, the kriged block estimates produce a smoothing effect resulting in underestimating high values and overestimating low values. The modified conditional simulation (MCS) methodology solves these problems by simulating the random field to preserve its mean and the variance structure. The simulation model is conditioned to reproduce the data at known sample points to minimize the variability between the simulated data and the true field data. In this study, the authors develop the MCS methodology to simulate ore reserve grades using the best linear unbiased estimation (BLUE) and the local average subdivision (LAS) techniques. The MCS methodology is applied to simulate block grades in a section of the Sabi Gold Project in Zimbabwe. The results are compared with the kriged estimates for this section. Analysis of the results shows that the MCS methodology reproduces the known sample grades with minimum estimation error of 0.001 while the estimation error associated with the kriged estimates is 1.104, a 100% efficiency of the MCS method over the kriging technique.     

Advances in Researches of the Effects of Grassland Vegetation on Soil Erosion in Loess Plateau

Soil erosion is a serious global environmental problem which limits the survival and development of human beings. In our country, due to the special physical geography and socio-economic conditions, soil erosion intensity is great, which is particularly prominent in Loess Plateau region. Therefore, preventing and controlling soil erosion, as well as reducing soil erosion in Loess Plateau have become the key to solving environmental problems in the region. Soil erosion on Loess Plateau is serious, and grassland vegetation has good effects on soil and water conservation, which can improve ecological environment well. After the implementation of the project about returning farmland to grassland on Loess Plateau, the ecological benefits mainly focused on soil and water conservation benefits, soil improvement benefits, water conservation benefits and species diversity benefits, etc. Grassland vegetation has an irreplaceable role in the construction of the ecological environment on Loess Plateau. Therefore, the role of grassland in preventing soil erosion has received more and more attention. Scholars have done lots of research involved in the relationship between grassland coverage and soil erosion, impacts of grassland on hydrodynamic parameters, effects of grassland on soil properties, reduction effects of grassland on runoff and sediment, and soil erosion process on grassland slope. However, there is little research on erosion effect induced by grassland cover. This paper mainly pointed out the following questions: First, grassland cover is influenced by many factors, but the relationship with soil erosion from the dynamic mechanism is rarely discussed; Second, there is no well-developed theory of overland flow erosion at present, which limits the study of hydrodynamic parameters on grassland slope; Third, establishment of mathematical model between grassland cover and soil resistance can accelerate the quantitative analysis of grassland influence on erosion; Fourth, comprehensive analysis of influencing factors on water reduction and sediment reduction effect on grassland are insufficient; Fifth, there are not many mechanisms to analyze the erosion process of grassland slope by using the hydrodynamic characteristics of slope; sixth, research results on grassland-induced erosion are mainly focused on leading to soil dry layer and we should continue to strengthen in the future. This paper summarized the previous results, and supplemented some studies about erosion caused by grassland, then pointed out the existing problems in current research and the areas that need to be strengthened in the future, aiming at reducing soil erosion on the Loess Plateau.     

Forward Modeling with Forced Neural Networks for Gravity Anomaly Profıle

In this paper, we introduce a new method called Forced Neural Network (FNN) to find the parameters of the object in geophysical section respect to gravity anomaly assuming the prismatic model. The aim of the geological modeling is to find the shape and location of underground structure, which cause the anomalies, in 2D cross section. At the first stage, we use one neuron to model the system and apply back propagation algorithm to find out the density difference. At the second level, quantization is applied to the density differences and mean square error of the system is computed. This process goes on until the mean square error of the system is small enough. First, we use FNN to two synthetic data, and then the Sivas–Gürün basin map in Turkey is chosen as a real data application. Anomaly values of the cross section, which is taken from the gravity anomaly map of Sivas–Gürün basin, are very close to those obtained from the proposed method.