任务驱动数据挖掘方法的提出及在低阻油层识别中的应用

为了解决数据挖掘算法的高效性、工具智能性与所获取知识无效性之间的矛盾,同时使所获取知识能有效地解决地层评价领域中的各种疑难问题,提出任务驱动数据挖掘方法。结合数据挖掘的概念和技术,阐述了任务驱动数据挖掘的概念和基本原理,分为建立数据仓库、数据预处理、选择特征子集、形成模型、模型评估、模型修正和模型发布等7个部分,是一个循环迭代的过程,直到构建能有效解决目标任务的预测模型为止。并以低阻油层的识别为例,详述整个分析处理过程,结合决策树形成的白盒模型和支持向量机构建的黑盒模型综合识别目标区的低阻油层,获得了准确率大于90%的识别效果。     

60年来大气中二氧化碳浓度数据的趋势方程研究

通过讨论已有的60年来大气中CO₂浓度数据的分布状态,采用趋势分析方法,给出了具体趋势方程形式.与冰芯分析或观测数据对比结果表明,趋势方程曲线与已有数据基本符合,随后初步给出了2010年至2016年间大气中CO₂浓度预测值.     

徐州轨道交通1号线工程地质灾害危险性预测评估

基于区域地质、水文地质及地质灾害方面的资料,并结合详细的野外调查、资料收集,采用定性、半定量的评估方法,对徐州市轨道交通1号线工程建设用地地质灾害危险性进行预测,工程项目可能引发或加剧及遭受的主要地质灾害类型有采空地面塌陷、岩溶地面塌陷、砂土液化等。针对评估结果提出地质灾害防治对策,以期为该项目防灾减灾、应急救援及类似工程地质灾害危险性评估提供参考。     

Power Maximization of A Point Absorber Wave Energy Converter Using Improved Model Predictive Control

This paper considers controlling and maximizing the absorbed power of wave energy converters for irregular waves. With respect to physical constraints of the system, a model predictive control is applied. Irregular waves’ behavior is predicted by Kalman filter method. Owing to the great influence of controller parameters on the absorbed power, these parameters are optimized by imperialist competitive algorithm. The results illustrate the method’s efficiency in maximizing the extracted power in the presence of unknown excitation force which should be predicted by Kalman filter.     

多步流量预测模型新方案的研究与应用

在分析1951-2001年潼关、花园口、兰州水文站年最大流量的变化规律基础上,提出了一个建立多步预测模型的新方案,该预测模型具有可以直接作多步预测和对异常极值有较好预测技巧等特点。将该预测模型用于1995-2001年多步独立样本的预测试验,取得了较好的预测效果,利用这一预测模型对未来直至2005年的年最大流量变化趋势进行了预测。     

Numerical Modeling of Coupled Geodynamical Processes and Its Role in Facilitating Predictive Ore Discovery: An Example from Tongling, China

Abstract. Facilitating more predictive ore discovery is becoming an increasingly challenging task for the resources geology research and exploration industry. The numerical geodynamic modeling, which uses numerical code to reproduce feedback coupling of geodynamic processes in the computer, is one of endeavors for forming sound predictive exploration strategy by thoroughly understanding the mineralization system and its controlling factors. The FLAC (Fast Lagrangian Analysis of Continua) is a two-dimensional finite-difference code for studying the mechano-thermo-hydrological behavior of a continuous porous medium after it reached equilibrium or steady plastic flow. We used the FLAC to simulate the syn-deformation cooling and fluid flowing after the intrusion was emplaced and solidified in the Fenghuangshan ore field, Tongling Cu district for guiding analysis of ore-controlling factors and selection of the exploration target area. The results of numerical computation demonstrate that the dilatant deformation zones along the contact of the intrusion control the fluid focus and the location and scale of ore bodies. Further more, the simulation results suggest a most prospective exploration area in the south high dilation zone, where the subsequent exploration supported the prediction and the test bore hole disclosed the high quality copper ore bodies in the target. This example has demonstrated the positive role of the numerical geodynamic modeling in facilitating predictive ore discovery.     

Biological assessment of rivers in the Manawatu‐Wanganui region of New Zealand using a predictive macroinvertebrate model

This study presents a river invertebrate and classification system (RIVPACS) type bioassessment methodology for the Manawatu‐Wanganui region of New Zealand. Aquatic macroinvertebrates and related physico‐chemical data were collected at 127 sites, with minimal human impacts (reference sites) in 2000. The reference sites were classified into five groups based on their macroinvertebrate data using TWINSPAN. These biotic groupings were then applied to their corresponding physico‐chemical data and discriminant functions were obtained to assign sites into the biotic groups using the physico‐chemical data. The discriminant functions correctly allocated 72% of the sites to the correct classification group using a jack‐knife validation. The probabilities from the discriminant functions were used to predict macroinvertebrate assemblages and these were compared with observed macroinvertebrate assemblages. The model was then used to assess the health of 29 test sites with known impacts. All test sites were assessed as impacted based on the 10th percentile of the reference data. To evaluate the temporal reliability of the model, data available for 11 sites sampled in 1997 and 2000 were run through the model. The results of this comparison showed little variation in O/E ratios over time and the two sites classed as impacted in 1997 were also classed as impacted in 2000.     

相控阵天气雷达数据压缩技术

相控阵天气雷达较新一代天气雷达在时空分辨率上有明显的优势。随着探测能力的提高,相控阵雷达数据量急剧增长,数据传输和存储问题凸显。现有天气雷达数据压缩算法可以减少传输和存储的数据量,但现有算法并未充分考虑相控阵雷达特点,数据压缩率有较大提升空间。本文提出时空预测的相控阵雷达数据压缩算法(PARDC),使用径向预测压缩数据相关冗余。使用CHATOR雷达数据对PARDC的算法性能进行评估,试验结果表明PARDC较通用的压缩算法压缩率性能大约提升了24%。     

遥感技术在腾冲芒棒盆地煤炭资源调查中的应用

以卫星遥感技术为主要手段，结合适量野外调查验证，综合分析了腾冲芒棒盆地聚煤作用及聚煤期后构造对煤层的改造作用，最终在盆地内圈定了两个遥感含煤远景区即橄榄寨——大崩嘎含煤远景区和弄岗——团田含煤远景区，为该区今后进一步煤炭资源勘探工作提供了靶区和基础资料。     

基于深度学习的山东大尹格庄金矿床深部三维预测模型

在隐伏矿体三维预测中,预测模型的准确性在很大程度上取决于找矿指标对矿化富集部位的指示性。然而,找矿指标容易受到找矿概念模型可靠性和成矿信息提取有效性限制,从而影响预测的准确性。论文以山东大尹格庄金矿隐伏矿体三维预测为例,基于深度学习方法,构建矿床深部隐伏矿体三维预测模型,旨在利用深度网络模型,学习获得对矿化具有显著指示性的找矿指标,提升三维预测的准确性。该方法将三维地质模型及其形态特征转换为适合卷积网络二维图像,采用卷积神经网络实现找矿指标的自动提取,并构建三维地质模型到矿化富集地段的定量关联。利用该方法建立了大尹格庄金矿的三维预测模型,经与几种人工建立找矿指标预测模型的对比分析,表明基于深度学习的预测模型较大地提升了预测准确性。