人工冻结梁构件中不同加筋作用的对比试验研究

考虑土层冻结构件中冻结管的加筋作用,对于增加冻结构件强度,减小变形量,提高人工冻结支护工程经济效益有重要作用。为研究冻结管对人工冻结构件的加筋作用,通过淤泥质粘土人工冻结梁构件三点弯试验,获得了各级载荷下构件的位移值。根据对冻土梁加筋与否,构件特定力学响应的比较,完成了冻结管加筋作用的量化分析。     

双星定位系统在中低轨卫星定轨中的应用

对双星定位系统在中低轨卫星轨道确定中的应用做了详细的分析，给出了在该系统的测量模式中低轨卫星定轨的几种方法。通过对这些方法进行的大量模拟计算和分析，结果表明作为双星定位系统的一种潜在功能，用其对中低轨卫星定轨是可行的。     

地铁隧道爆破施工影响的邻近居民楼质点峰值振速研究

地铁隧道爆破施工时严重威胁邻近居民楼的结构安全,因此研究邻近居民楼爆破峰值振速是非常必要的。文章基于群孔荷载简化模型、柱面波理论及传统的应力波衰减理论,在考虑高程效应影响下推导质点峰值振速公式。依托乌鲁木齐地铁一号线工程,对一邻近六层居民楼的爆破施工响应进行监测,并将实测质点峰值振速、提出的修正公式预测振速和传统萨道夫公式预测的振速进行对比分析。研究结果表明:采用修正公式预测的质点峰值振速与现场实测值的分布规律一致,呈现出随着高程增大而逐渐增大的趋势;而且现场实测值与修正公式预测值相对误差较小,表明考虑高程因子的修正公式是可行的。     

地质领域机器学习、深度学习及实现语言

地质大数据正在以指数形式增长。只有发展智能数据处理方法才有可能追上大数据的超常增长。机器学习是人工智能的核心,是使计算机具有智能的根本途径。机器学习已成为地质大数据研究的前沿热点,它将让地质大数据插上翅膀,并因此改变地质。机器学习是一个源于数据的模型的训练过程,最终给出一个面向某种性能度量的决策。深度学习是机器学习研究中的一个重要子类,它通过构建具有很多隐层的机器学习模型和海量的训练数据,来学习更有用的特征,从而最终提升分类或预测的准确性。卷积神经网络算法是最为常用的一种深度学习算法之一,它广泛用于图像识别和语音分析等。Python语言在科学领域的地位占据着越来越重要。其下的Scikit-Learn是一个机器学习相关的库,提供有数据预处理、分类、回归、聚类、预测、模型分析等算法。Keras是一个基于Theano/Tensorflow的深度学习库,可以应用来搭建简洁的人工神经网络。     

凡纳对虾人工繁殖的初步研究

研究了海水比重的变化对凡纳对虾亲虾人工繁殖活动的影响及换水对亲虾交配率的影响。试验采用剪眼柄、强化培育、控光、换水等技术措施 ,将凡纳对虾培育成熟并使之自然交配产卵 ,孵化出无节幼体 ;亲虾的平均产卵量可达 1 6万粒左右 ,卵的受精孵化率可达 80 %以上。结果表明 :适时换水可将亲虾的交配率从 4 6 %提高到 79% ;凡纳对虾人工繁殖的适宜海水相对密度为 1 0 1 6以上。     

Mapping of hard rock aquifer system and artificial recharge zonation through remote sensing and GIS approach in parts of Perambalur District of Tamil Nadu,India

Mapping of hard rock aquifer system and artificial recharge zonation were carried out in an area of 325 km2 in parts of the Perambalur District, Tamil Nadu, India. This district has been declared as one of the over-exploited regions in Tamil Nadu by the Central Groundwater Board. To raise the groundwater level, suitable recharge zones were identified and artificial recharge structures are suggested using geomatics technology in the present study. To this end, various thematic maps concerning lithology, soil, geomorphology, land use, land cover, slope, lineament, lineament density, drainage, drainage density and groundwater depth level were prepared. Fissile hornblende gneiss (244 km2) covered most of the study area followed by charnockites (68 km2). Structural hills and rocky pediments characterize the major geomorphological features in the targeted area, and are followed by deep moderated pediments. The area is mostly used as crop and fallow land, followed by scrub land and deciduous forest. In the study area, the slopes are predominantly very gentle (142 km2) and nearly level (66 km2) ones. Besides, Groundwater level data of 58 wells have been generated, in which the minimum and maximum depth were 3 and 28 m respectively. Integration under the GIS environment has been carried out using all the thematic layers to identify the groundwater prospect zone through the introduction of weight and rank methods. Integrated output performances were classified into very poor, poor, moderate, good and excellent categories. All these classes were further divided into two groups as suitable and non-suitable area for the selection of recharge sites. Hard rock fractures were mapped as lineaments from satellite images, and besides that, rose diagram was also generated to find out the trend of the fracture. Furthermore, fracture data of 146 numbers have been collected using Brunton compass to generate rose diagram and were correlated with the rose diagram derived from lineaments. The present study significantly brought up a few areas such as Ammapalayam, Melapuliyur, Senjeri and around Siruvachur for artificial recharge.     

基于HYDRUS-2D的渗滤池回灌地下水数值模拟研究

人工回灌是提高水资源利用水平的重要工程措施。深入研究渗滤池回灌过程机理,基于Hydrus-2D,结合具体渗滤池回灌算例,建立渗滤池回灌的二维饱和-非饱和渗流数值模型。模拟结果表明：回灌水入渗形成的＂湿润峰＂逐渐向饱和带移动,同时饱和带的浸润线逐渐上升,形成不断扩大的饱和区,直到形成一个连续的饱和带。回灌水出流速率经过了迅速增大到一峰值后快速减小,然后又缓慢变小的过程,回灌水入渗速率随时间变化的这一特征与Green-Ampt入渗模型的结论是一致的。     

宁芜盆地白象山矿区人工神经网络 三维成矿定位预测研究

本文基于三维地质环境,综合白象山矿区积累的地质资料和物探成果,首先开展三维地质建模工作,详细刻画了白象山矿区的三维地质结构;在三维地质模型基础上,利用三维空间分析手段对三维控矿因素进行定量挖掘,提取了多种三维控矿因素;最后采用人工神经网络方法进行三维成矿定位预测。预测结果显示,人工神经网络三维成矿定位预测能很好的定位出已知矿体,同时显示,在已知矿体北部及东部的深边部具有较高的成矿概率,可作为开展进一步找矿勘探的靶区。因此,人工神经网络三维成矿定位预测对于白象山矿区的应用是有效的,可服务于新老矿区的深边部三维成矿定位预测,同时可为隐伏矿、盲矿的成矿预测和优选靶区提供定量、定位新的方法和途径。     

DEM–SPH simulation of rock blasting

A numerical model is proposed for the simulation of rock blasting. A bonded particle system is utilized to mimic the behavior of rock. The particles interact at the contact points through normal and shear springs to simulate rock elasticity. To withstand the deviatoric stresses, the particles are glued to each other. If the applied force exceeds the contact strength, local failure occurs and microcracks are developed in the synthetic rock. For simulation of gas flow, the smooth particle hydrodynamic method is implemented. The interaction of gas particles with the rock grains is assumed to follow a perfect plastic collision model in which the initial momentum of the colliding particles is preserved. A detailed examination of the interaction of gas with blast hole is investigated. It is shown that the proposed hybrid model is capable of simulating the induced shock waves in the gas together with wave propagation in the rock material. The model successfully mimics crack propagation in rock. In particular, the crushed zone around the borehole, radial cracks, and surface spalling are all captured successfully. The results of numerical analysis suggest that gas–rock interaction can, in fact, generate a few successive compressive waves in the rock specimen, causing further extension of radial cracks with time as the weaker secondary and tertiary waves interact with the crack tips.     

水下爆破动水压力预测及衰减规律研究

本文基于传统的Cole动水压力预测公式,采用量纲分析法建立了考虑多影响因素的动水压力相似准数方程,并在此基础上进行现场监测试验,依据试验数据进行非线性回归运算,得出水下爆破动水压力预测模型公式相关参数。将得到的参数回代入Cole公式和新建模型公式进行预测误差估计和准确度评价分析,二者相对误差范围分别为0.64%～17.11%和0.08%～6.53%,平均相对误差分别为7.31%和3.33%,证明了新建模型公式对动水压力峰值的预测相比传统Cole公式更接近于实际。同时依据修正得到的模型公式对动水压力衰减规律进行分析,结果表明：动水压力峰值随着单段最大药量的增加而增大,随着传播至覆盖层表面振速峰值的增大而增大,随着距爆心水平距离的增大而不断衰减。