Experimental study on mitigating wind erosion of calcareous desert sand using spray method for microbially induced calcium carbonate precipitation

Wind erosion is one of the significant natural calamities worldwide, which degrades around one-third of global land. The eroded and suspended soil particles in the environment may cause health hazards, i.e. allergies and respiratory diseases, due to the presence of harmful contaminants, bacteria, and pollens. The present study evaluates the feasibility of microbially induced calcium carbonate precipitation (MICP) technique to mitigate wind-induced erosion of calcareous desert sand (Thar desert of Rajasthan province in India). The temperature during biotreatment was kept at 36 °C to stimulate the average temperature of the Thar desert. The spray method was used for bioaugmentation of Sporosarcina (S.) pasteurii and further treatment using chemical solutions. The chemical solution of 0.25 pore volume was sprayed continuously up to 5 d, 10 d, 15 d, and 20 d, using two different concentration ratios of urea and calcium chloride dihydrate viz 2:1 and 1:1. The biotreated samples were subjected to erosion testing (in the wind tunnel) at different wind speeds of 10 m/s, 20 m/s, and 30 m/s. The unconfined compressive strength of the biocemented crust was measured using a pocket penetrometer. The variation in calcite precipitation and microstructure (including the presence of crystalline minerals) of untreated as well as biotreated sand samples were determined through calcimeter, scanning electron microscope (SEM), and energy-dispersive X-ray spectroscope (EDX). The results demonstrated that the erosion of untreated sand increases with an increase in wind speeds. When compared to untreated sand, a lower erosion was observed in all biocemented sand samples, irrespective of treatment condition and wind speed. It was observed that the sample treated with 1:1 cementation solution for up to 5 d, was found to effectively resist erosion at a wind speed of 10 m/s. Moreover, a significant erosion resistance was ascertained in 15 d and 20 d treated samples at higher wind speeds. The calcite content percentage, thickness of crust, bulk density, and surface strength of biocemented sand were enhanced with the increase in treatment duration. The 1:1 concentration ratio of cementation solution was found effective in improving crust thickness and surface strength as compared to 2:1 concentration ratio of cementation solution. The calcite crystals formation was observed in SEM analysis and calcium peaks were observed in EDX analysis for biotreated sand.     

基于卷积神经网络的5G蜂窝网络无线定位方法

5G蜂窝网络发展迅猛，其覆盖面积将逐渐增大，因此使用5G蜂窝网络进行定位是有研究潜力的研究方向。本文提出一种新的深度学习技术来实现高效、高精度和低占用的定位，以代替传统指纹定位过程中繁重的指纹库生成以及距离计算。该方法建立了一个特殊的卷积神经网络，并根据5G天线信号的接收信号强度指示、相位和到达角等特征量，选择合适的输入数据格式构造样本组建训练集，对该卷积神经网络进行训练。训练得到的卷积神经网络可以替代指纹定位中的庞大指纹库，非常有利于直接在5G移动设备端实现定位。虽然卷积神经网络在训练过程中需要大量时间，但在训练完毕后直接进行分类定位的速度非常快，可以保障定位实现的实时性。本文所实现的卷积神经网络权重与偏置所占内存不到0.5 MB，且能够在实际应用环境中以95%的定位准确率以及0.1 m的平均定位精度实现高精度定位。     

Study on the difference of dispersion behavior between hydrogen and methane in utility tunnel

Compared to liquid/gas hydrogen tank, the pipeline is an economical way for hydrogen transportation. With the quick development of utility tunnel in China, hydrogen pipeline enters the gas compartment can be expected soon. However, all the safety requirements of the gas compartment in the current standards are designed for natural gas, and the applicability for hydrogen is unknown. Therefore, a series of studies were started to investigate the safety of hydrogen in utility tunnel. In this work, a real utility tunnel locates at Shanghai was selected as the physical object. A 3D numerical model was built and successfully validated by a scaled tunnel test. The model has the maximum deviation of +9.5%. After that, a comparatively study of the dispersion behavior of CH₄ and H₂ was conducted. The assumed scenario was a 20 mm small-hole leaks with gauge pressure of 1.0 MPa in the middle of the tunnel. Numerical results shown that, H₂ has a larger dispersion velocity and higher concentration, and is more dangerous compared to CH₄. The current emergency ventilation strategy of air change rate of 12 times/h is not effective enough to dilute the H₂ flammable cloud. The alarm time of the testing points shown strong linear law. There was a sharp variation in the range of 20%–100% LFL (Lower Flammable Limit), so the alarm strategy in the tunnel standards is too ideal for both CH₄ and H₂. The numerical results in the present study could provide a guidance for the design and safety management of the hydrogen tunnel.     

基准平面垂直断面法在爆破漏斗试验中的应用

系统阐述了基准平面垂直断面法在爆破漏斗试验中测量爆破漏斗体积的基本原理,并将隧道激光断面仪应用于金厂河矿1 750 m水平15^#采场底部切割巷道爆破漏斗试验爆破漏斗体积测量中。通过与传统体重法等计算法所得漏斗体积分析比较,结果表明基于隧道激光断面仪与3D Mine软件分析的基准平面垂直断面法实用性强、操作方便、结果直观可靠,达到试验预期目的。     

基于阶段注意力机制的电力线提取算法

电力系统维护是电力系统稳定运行的重要保障，应用智能算法的无人机电力巡检则为电力系统维护提供便捷。电力线提取是自主电力巡检以及保障飞行器低空飞行安全的关键技术，结合深度学习理论进行电力线提取是电力巡检的重要突破点。本文将深度学习方法用于电力线提取任务，结合电力线图像特点嵌入改进的图像输入策略和注意力模块，提出一种基于阶段注意力机制的电力线提取模型（SA-Unet）。本文提出的SA-Unet模型编码阶段采用阶段输入融合策略（Stage input fusion strategy， SIFS），充分利用图像的多尺度信息减少空间位置信息丢失。解码阶段通过嵌入阶段注意力模块（Stage attention module，SAM）聚焦电力线特征，从大量信息中快速筛选出高价值信息。实验结果表明，该方法在复杂背景的多场景中具有良好的性能。     

胶带斜井在冶金矿山深井开拓中的应用探讨

针对某在建矿山深井开拓方案探索需求，通过对胶带运输系统发展应用情况介绍和胶带运输系统运输特性影响因素分析，并结合深井开拓胶带斜井方案与竖井加胶带斜井方案投资经济性比较，以及对胶带运输系统在运行过程中关键问题的分析和讨论，得出随着胶带运输技术的不断发展，在大规模深井开拓中，胶带斜井开拓将大有可为，为冶金矿山深井开拓方案选择提供决策支持。     

用于厚屏蔽小探测器的蒙特卡罗模拟减方差方法研究

反应堆屏蔽计算中经常出现厚屏蔽、小探测器问题，常规蒙特卡罗方法难以有效解决。基于自动重要抽样（AIS）方法，本文提出了小探测器自动重要抽样(SDAIS)方法，并针对小探测器问题，优化了AIS方法的虚粒子赌分裂算法。该方法在自主开发的蒙特卡罗屏蔽程序MCShield上进行了实现。使用NUREG/CR-6115 PWR基准题验证该方法的正确性和计算效率。结果表明，SDAIS方法可有效地解决厚屏蔽小探测器问题，相比AIS方法及传统的重要性方法，计算效率提升1~2个量级。     

结冰风洞校准标准研究

本文按结冰风洞制冷方式、试验对象、用途性质和试验段尺寸归纳总结了结冰风洞的类型,指出校准标准主要针对的结冰风洞类型;阐述结冰风洞空气动力流场校准可以依据的3个国内外标准;介绍SAE结冰风洞校准标准中云雾场校准的主要内容;分析研究相关标准的特点和存在的不足;提出未来我国制定结冰风洞校准标准的建议。     

基于深度信念网络的列车故障音频诊断方法

铁路在交通运输行业有着举足轻重的地位，一旦列车发生故障将会导致严重的生命财产损失。由于列车发生故障的概率相对较低，因此难以捕获列车的故障样本。针对上述问题，提出了一种无监督学习的列车故障识别方法，通过检测列车音频信号来识别列车故障。该方法基于深度信念网络(DBN)，利用小波包分解提取检测信号的特征向量并将其作为DBN的输入，待网络充分训练后，由训练好的DBN识别当前列车的运行状况。现场监测实验结果表明，该方法能够在无监督的条件下有效识别列车故障，保障了列车的运行安全。     

基于深度学习的秀丽隐杆线虫显微图像分割方法

利用计算机实现自动、准确的秀丽隐杆线虫(C.elegans)的各项形态学参数分析,至关重要的是从显微图像上分割出线虫体态,但由于显微镜下的图像噪声较多,线虫边缘像素与周围环境相似,而且线虫的体态具有鞭毛和其他附着物需要分离,多方面因素导致设计一个鲁棒性的C.elegans分割算法仍然面临着挑战。针对这些问题,提出了一种基于深度学习的线虫分割方法,通过训练掩模区域卷积神经网络(Mask R-CNN)学习线虫形态特征实现自动分割。首先,通过改进多级特征池化将高级语义特征与低级边缘特征融合,结合大幅度软最大损失(LMSL)损失算法改进损失计算;然后,改进非极大值抑制;最后,引入全连接融合分支等方法对分割结果进行进一步优化。实验结果表明,相比原始的Mask R-CNN,该方法平均精确率(AP)提升了4.3个百分点,平均交并比(mIOU)提升了4个百分点。表明所提出的深度学习分割方法能够有效提高分割准确率,在显微图像中更加精确地分割出线虫体。