The idea and project of the “Medium－Scale Experiment Field for Earthquake Prediction”──Research on observations and applications of mining earthquake in Mentougou Coal Mine

A brief account of the development of the research on mining earthquakes and the general situation of the Mentougou Coal Mine medium scale experiment field for earthquake prediction and the project of monitor and prediction is given. The differences of waveforms between mining earthquakes and natural earthquakes is discussed. The magnitude frequency distribution of the 79 000 mining earthquakes of over M L1.0 from 1984 to 1995 is summarized . Finally, taking PH and PV, the principal compressive stress components of the focal mechanism of the mining earthquakes, as the criteria, analyses the stress background of the 12 large mining earthquakes.

Mercury-Free Gold Extraction Using Borax for Small-Scale Gold Miners

Small-scale gold mining is the largest anthropogenic contributor of mercury pollution on planet Earth. The miners grind gold ore together with mercury in order to recover the fine gold grains. The gold amalgam is burned whereby the mercury evaporates and gold is left behind. This processing extracts only a fraction of the gold in the ore and the released mercury causes serious global health and environmental problems. However, a mercury-free method described below has proved to extract up to twice as much gold without need of investment in new processing equipment and without longer processing time. The method has successfully been tested in the Philippines, Indonesia, Tanzania, Bolivia and Zimbabwe.

Fire behavior of mining vehicles in underground hard rock mines

The results from a number of investigations and fire experiments are presented and analyzed in order to characterize the fire behavior of mining vehicles in underground hard rock mines. The analysis also includes fire safety and fire protection measures with respect to the mining vehicle fire behavior.Earlier studies on fires in underground hard rock mines have shown that vehicles or mobile equipment are the dominant sources of fire. A better knowledge about the fire behavior of vehicles in underground hard rock mines is therefore needed. During the analysis the direction and flow rate of the ventilation in a drift was found to have a significant impact on the fire behavior, causing for example flame tilt with rapid fire spread. The shielded sections of a vehicle will be less affected by the ventilation flow resulting in for example a decreased flame spread. It was also found that spray fires may result in considerable heat release rate but are generally of shorter duration and will not make any significant contributions to the overall heat release rate of the fully developed vehicle fire. The fire duration of a loader tire from a full-scale fire experiment was found to be at least 200 min and will largely determine the total fire duration of the vehicle. A different scenario with different conditions with for example a slower flame spread resulted in an even longer fire duration. The radiative and convective fraction will be a key factor when determining the heat transfer mechanisms involved in a fire and will vary from material to material.Calculations show that the radiative fraction of the tire fires on two mining vehicles is significantly lower than found in earlier experiments. The design and construction of the mining vehicle will have an important impact on the fire behavior and could possibly mitigate the consequences of a fire and allow fire personnel to extinguish a fire that otherwise would have had a too high heat release rate.

Whole-mine subsidence over tabular deposits and related seismicity

The challenge of estimating mine-wide subsidence and linkages to seismicity over tabular deposits is addressed by a special finite element technique(dual node–dual mesh). Subsidence and mine-induced seismicity begins near the face when caving occurs and propagates to the surface as extraction reaches a critical extent. Thus, the challenge is to obtain details at the face at the meter scale and also at the surface over the whole mine at the kilometer scale. Interactions between old and new sections of a mine are automatically taken into account with this technique. The finite element method is well established technology based on fundamentals of physical laws, kinematics and material laws. With this technique, no empirical ‘‘scaling" or fitting computer output by input data ‘‘adjustment" to mine measurements is necessary. Capability is demonstrated for doing practical whole-mine subsidence analysis from first principles. Mine-induced seismicity is shown to correlate well with face advance and element failure.

Heavy Metal Pollution in Soil and Water in Some Selected Towns in Dunkwa-on-Offin District in the Central Region of Ghana as a Result of Small Scale Gold Mining

Illegal small scale gold mining popularly called “Galamsey” in our local communities is on the increase. This has led to concerns about the level of environmental pollution resulting from these mining activities. This work was conducted to determine the level of heavy metal contamination in the environment due to the activities of the small scale miners. This paper discusses the concentrations of some selected heavy metals—Hg, Pb, and Cu which were measured in 14 sampling sites in Dunkwa-on-Offin in the Central Region of Ghana, known for these activities. The heavy metal concentrations have been investigated for soil and water samples in the selected towns and compared with the relevant guidelines of the Environmental Protection Agency. The concentration of heavy metals was measured by using AAS. In most locations, the concentration for the investigated heavy metals far exceeded the concentration admitted by the guidelines. The mean concentration of Lead was 95.13 mg/Kg for soil and 190.27 mg/L in water;Copper was 63.26 mg/Kg in soil and 75.92 mg/L in water while Mercury was 140.87 ug/Kg in soil and 211.31 mg/L in water. The mean recorded concentrations in the sensitive areas exceeded greatly. Hence the levels of heavy metal contamination have spread beyond control.

Small Scale Mining in Tanzania Toward Sustainable Use(Prospects,Opportunities and Challenges for Environmental Management)

Although small scale mining management is a complex task,many countries in the world have managed to develop mining management regimes that can improve the economic and environmental efficiency of the small scale mining sector.This paper attempts to explore the mental picture of opportunities and challenges of the environmental management in small scale mining in Tanzania focusing on

The Need to Regularise Activities of Illegal Small-Scale Mining in Ghana: A Focus on the Tarkwa-Dunkwa Highway

Southwest Ghana is a major gold producing region. The current high gold price has attracted hundreds of unemployed youth to undertake small-scale mining (SSM). Most of these miners operate illegally even though the SSM law (PNDCL) 218 of 1989 and Act 703 of 2006 define the procedures required for their operation. Some miners have brought their activities to a segment of the western highway that links southwest to central Ghana with serious environmental consequences envisaged. This paper argues that the laws that regulate SSM do not consider the fundamental set-up and concerns of the small-scale miner, hence its inability to be effective. It is therefore proposed that the present requirement that a minimum of 21 hectares is necessary before land can be registered needs re-examination. Secondly, government needs to explore parcels of land and designate the workable areas to miners under a well structured scheme that will also educate these miners about safe and healthy mining methods.

MapReduce模型在大规模数据并行挖掘中的应用

MapReduce并行编程模型通过定义良好的接口和运行支持库,能够自动并行执行大规模计算任务,隐藏底层实现细节,降低并行编程的难度。系统阐述MapReduce的基本工作原理及其工作流程,以TeraSort算法为例,针对其存在的问题,提出动态数据分区和数据压缩等优化建议。研究成果表明,优化后的TeraSort算法能够显著缩短数据处理时间,优化系统的吞吐量,并改善资源分配的均衡性。

基于多尺度特征融合和注意力机制的矿区道路障碍检测

为解决复杂露天矿区背景下的行车障碍检测问题,提出一种基于改进跨尺度特征融合的矿区道路障碍检测算法。首先,针对原始矿山数据集中小目标样本类别不平衡的问题,采用一种基于几何变换和加权泊松融合的数据增强方法扩大样本数量;其次,在特征提取阶段提出更适用于障碍检测的跨阶段连接网络,以增大检测尺度,提高算法对小目标特征的学习能力;然后,在特征融合阶段提出基于3D无参注意力(SimAM)和去权重的双向特征融合金字塔网络(Bi-FPN),通过扩大预测特征图和特征感受野,提升多尺度检测性能;最后,针对训练中样本不均衡和障碍物边界框定位不精准问题,引入质量焦点损失函数(QFL)和可扩展的交并比损失函数(SIoU),将分类得分与位置的质量预测结合,提高对密集遮挡目标的定位精度。结果表明:改进方法能有效识别复杂背景下露天矿区非结构化道路障碍物,在实际应用中,检测精度达到91.88%,检测速度达到68.7帧/s,相较于主流检测方法有着更好的小目标和多尺度检测性能,可满足露天矿区无人矿卡行进中的障碍安全检测要求。

利用手持激光点云的矿区自然林树干提取

为了定量刻画矿区复杂环境下自然林树木的三维结构信息,提出一种利用手持激光扫描仪的矿区自然林树干点云识别方法。借助手持激光扫描仪对矿区自然林进行扫描,在滤除地面点的基础上,引入多尺度垂直度特征进行矿区自然林树干识别,利用八叉树叶节点聚类方法对识别的树干进一步提取。通过矿区实测数据进行实验,结果表明,所提出的方法能够实现矿区自然林树干的识别,对矿区复杂环境下自然林树木的三维结构信息定量刻画、厘清矿区植被生长状况具有一定的参考价值。

论地下矿山大型机械化开采安全管理技术的应用

地下矿山的大型机械化开采具有高危险性,为提高生产效率并降低安全风险,先进的安全管理技术成为必不可少的环节。本文以矿山为背景,探讨了安全管理技术在大型机械化开采中的应用。通过详细分析数据分析与预警系统、智能监控与控制系统在案例中的具体作用,以及安全培训与意识提升的重要性,揭示了这些技术在提高安全性的同时实现了与生产效率的平衡。成功案例的深入分析为未来的改进提供了有益的教训,强调了综合运用不同技术和持续改进的重要性。

束状孔分层落矿采矿方法研究

针对球形药包分层崩落采矿法落矿分层厚度小、采场生产能力较低的问题,基于束状孔爆破理论,研发束状孔分层落矿采矿方法。束状孔爆破具有增强破岩作用的爆破效应。4~5个炮孔组成的束状孔,采用当量球形药包装药,爆破分层高度为单孔球形药包爆破的2~3倍。束状孔分层落矿采矿方法采用束状深孔当量球形药包漏斗爆破,束状炮孔间距7~8 m,采用普通工业炸药爆破。每次分层爆破落矿高度5~8 m,贫化率8%,损失率10%,大块率<5%,采场生产能力1 800~2 400 t/d。与单孔球形药包爆破VCR法比较,大块率降低10%~20%,采场生产能力提高1倍以上。束状孔分层落矿采矿方法是一种适用于厚大矿体、中厚急倾斜矿体的大规模高效开采方法。

基于多尺度特征匹配的矿井排水系统故障诊断算法

矿井排水系统的高效运行对于煤矿生产具有重要意义,然而,故障的发生会导致排水系统运行异常,进而影响生产安全和效率。提出了一种基于多尺度特征匹配的矿井排水系统故障诊断算法。该算法首先通过传感器获取到实时的矿井排水系统数据,并对其进行预处理和特征提取。然后,采用多尺度特征匹配方法,将数据特征映射到不同的尺度空间中。在每个尺度空间中,通过构建特征匹配模型,将当前数据与已知的正常运行数据进行比对,以判断是否存在故障。为了提高故障诊断的准确性和鲁棒性,引入集成学习方法,通过集成多个尺度空间的特征匹配结果,并利用投票机制进行综合决策,最终确定矿井排水系统的故障类型和位置。研究表明:所提算法在矿井排水系统故障诊断方面具有良好性能,与传统方法相比,该算法能够高效地检测故障,有助于及时准确地发现和解决矿井排水系统问题,对于确保矿井安全生产具有一定的意义。

复杂条件超大规模连续盘区多中段开采应力扰动规律研究

以某超大规模开采铁矿为研究对象,模拟了矿区超大规模连续盘区多中段充填采矿过程,重点对回采过程中的围岩扰动应力演化规律进行了分析。结果显示多中段开采的扰动叠加效应对26 m矿柱稳定性影响显著。两中段之间存在卸压现象,深部采区-580 m水平以下应力集中明显。研究结果可供连续盘区超大规模开采、深井矿山卸压开采等借鉴参考。

姿势自适应权重更新策略及尺度回归的煤矿井下目标跟踪算法

随着计算机技术的飞速发展,煤矿井下智能工作面的建设取得了巨大进步。目标跟踪作为井下智能工作面建设的一个重要环节,在保证作业人员安全,提升开采效率等方面发挥着重要作用。但是煤矿井下光照条件较差,当目标有较大外观变化和遮挡时,算法的跟踪预测框将不能准确跟踪目标。为了解决上述问题,设计出姿势自适应权重更新策略,该策略能计算出不同帧与模板帧之间目标外观姿势的变化程度,该变化程度作为目标估计网络调制向量更新的依据,动态地更新调制向量。通过该更新策略,使得算法对于目标尺度变化较大物体的跟踪精度得到提升。同时设计的中心点尺度回归损失函数,将中心点距离和高宽损失引入损失函数,使得目标框回归更加精准。

基于数据挖掘算法的精准营销方法

以标准化流程为基础,抽取营销数据中的关键性指标;采用无量纲转换方式,对关键指标的精准营销尺度进行统一转化;通过数据挖掘算法标定尺度关系,基于数据之间的逻辑关系实现精准营销。实验结果表明:以不同类型用户作为营销变量,通过新方法可以实现不同需求的精准匹配,具有应用价值。

基于无监督学习的异质网络多尺度离群点挖掘研究

现有的异质网络多尺度离群点挖掘算法忽略了数据点之间的顺序关系,无法充分利用数据点在异质网络中的排列顺序信息,从而导致聚类精度下降。对此,提出一种基于无监督学习的异质网络多尺度离群点挖掘方法,对异质网络的多节点、多边特点进行分析。利用季节-趋势时序分解法提取异质网络数据特征。根据数据特征,结合K-means聚类算法与排序算法,将数据点的排序信息添加至聚类过程中,以实现对异质网络数据离群点的挖掘。实验结果表明,利用该方法进行网络数据节点聚类的准确率均能达到80%以上;并且实现了多尺度离群点挖掘后,能够精准地识别出离群点,为后续的网络通信维护提供了良好的保障。

机载激光雷达技术在矿山大比例尺地形图测绘中的应用

针对常规无人机挂载可见光相机航测,通过空中三角测量得到地面三维坐标信息的生成过程,受制于地面植被覆盖密度高、地形起伏大,可见光航测相机往往只能得到植被表面的高程数据,即获取到DSM(Digital Surface Model,数字表面模型)数据,无法获得地面高程数据,即DEM(Digital Elevation Model,数字高程模型)数据,无法满足大比例尺地形图生成的需要。利用无人机挂载激光雷达设备,激光雷达可以穿透地表植被,直接测量到真实地表,将点云分类后制作DEM,用于大比例尺地形图生成中,能够解决矿区井田落差大、植被茂密情况下,传统测量速度慢、可见光航测技术精度无法满足要求的技术难题。

小样本机器学习下数据多尺度挖掘算法设计

数据多尺度挖掘是指在数据挖掘过程中考虑不同尺度的数据信息,由于数据中存在大量特征,为了提高信息处理效率,且更精准地划分数据类型,提出一种小样本机器学习算法下数据多尺度挖掘方法。将使用者的部分动作抽象,建立网站页面会话用于学习不同的事件内容,获得小样本数据信息。通过复值函数构建Hibert空间,计算出样本元素再生核,提取小样本数据特征;利用特征向量构造特征矩阵调节数据间平衡性,得到数据相对熵。建立多尺度信息数据库,使用机器学习下逻辑回归离散化数据特征值,挖掘复杂项集指标的支持度,实现精准的数据多尺度挖掘。通过实验证明,所提方法数据分类效果好,挖掘准确率高,耗费时间少。