Performance Evaluation of Low-Carbon and Clean Transformation of China’s Coal Economy

In China,the oversupply of coal occurred in 2009,and from that year onwards,China’s coal economy began a low-carbon and clean transformation.Evaluating transformation performance is the research goal of this paper.The data collection for this paper includes data on deep processing of Chinese coal products from 2009 to 2020,as well as data on asset structure evolution and financial performance of 34 listed companies in the Chinese coal mining.Entropy value method is used to calculate the entropy value of low-carbon transformation,and the regression analysis is used to study the performance of cleaner transformation,the conclusion is as follows:(1)From 2009 to 2020,in China’s total energy consumption,coal consumption accounted for 71.6%in 2009 and 56.8%in 2020,the goals set by the state have been achieved.(2)The national goal of reducing the proportion of coal consumption and reducing carbon emissions has forced the transformation of deep processing of coal products.The transformation of coal enterprises towards low-carbon and clean production has achieved remarkable results.(3)From 2009 to 2020,the non coal industry income of 34 listed companies in China’s coal mining industry increased by 8.21%annually.At the same time,the asset structure was adjusted,and nearly 80%of the asset structure evolution showed an orderly development trend.(4)The regression analysis results show that the entropy value of coal deep processing products and the entropy value of asset structure adjustment are significantly related to transformation performance.The paper proposes to summarize the successful experience of China’s coal energy economic transformation,lay a foundation for achieving the carbon peak and carbon neutral goals in the future,further increase the intensity of coal deep processing,increase the proportion of clean energy in total energy consumption,and strive to control asset operation towards the goal of increasing the proportion of non coal industry income.

Coupling Effects on Gold Mineralization of Deep and Shallow Structures in the Northwestern Jiaodong Peninsula, Eastern China

For understanding the possible deep-seated processes and geodynamic constrains on gold mineralization, comprehensive physicochemical and geochemical studies of gold mineralization have been undertaken within the paleo-lithosphere framework during the metailogenic epoch from the northwestern part of the Jiaodong Peninsula in this paper. A general image of the paleo-crust has been remained although it has been superimposed and reformed by post-metailogenic tectonic movements. The gold ore deposits occur usually in local uplifts and gradient belts featuring a turn from steep to gentle in granite-metamorphic contact zones, relative uplifts of gradient zones of the Curier isothermal interfaces, depressions of the Moho discontinuity and areas where depth contours are cut by isotherms perpendicularly. Gold mineralization and lithogenesis are characterized by high temperature, low pressure and high strength of thermal flux. The depth of mineralization ranges from 0.8 to 4.5 km. The depth of the top interface of the granitic complex in the metallogenic epoch is about 3 km. There is a low-velocity layer （LVL） at the bottom of the upper crust with a depth close to 19.5 km, which may be a detachment belt in the crust. The appearance of the LVL indicates the existence of paleo-hyperthermal fluid or relics of molten magma chambers, which reflects partial melting within the crust during the diagenetic and metallogenic epochs and the superposition effects of strike-slip shearing of the Taulu fault zone. The subsidence of the Moho is probably attributed to the coupling process of the NW-SE continental collision between North China and the Yangtze Block and the strike-slip movement of the Tanlu fault accompanied with underplating of mantle magma in the northwestern part of the Jiaodong Peninsula. The underplating of mantle magma may result in partial melting and make granite magma transfer upwards. This is favorable for the migration of metallogenic materials from deep to shallow to be enriched to form deposits. Coupling interactions between the strike-slip of the Taulu fault, the underplating of mantle magma, partial melting within the crust, and hyperthermal fluid, etc. may be the important factors controlling the gold mineralization and spatial structures in the metailogenic system.

Study on coupling between deep and shallow structures of Xingtai area and some significant questions

In the light of results from study on coupling between deep and shallow structures in Xingtai earthquake area during the 'Ninth Five-Year Plan' period and other previous results from deep seismic refraction/reflection and seismic prospecting of petroleum, we infer that there exist a series of shallow faults in the upper crust above the 8 km-deep detachment surface in Xingtai macroseismic focal region, where none of the faults, including Aixinzhuang fault reaches the Quaternary stratum, except that the Xinhe fault cuts through the mid-Pleistocene formation upwards. Aixinzhuang fault and other faults extend downwards into Xinhe fault whereas the Xinhe listric fault stretches downwards at a low dip angle into the detachment surface. The abyssal fault with high dip angle under the detachment surface cutting through the middle and lower crust to Moho is the causative fault for the large Xingtai earthquake, whose dislocation can cause strong earthquakes, shallow fault activity and the motion of surface material. The shallow faults in the upper crust are not causative faults for strong earthquakes, although they may be active faults. The existence of the detachment surface brings about a special relationship between shallow and deep structures, i.e. they are relatively independent of each other and have effects on each other It not only transmits partial energy and deformation between the upper and lower crust,but also has a certain decoupling effect. Finally we conclude that active faults do not necessarily reach the latest stratum, and the age of uppermost faulted stratum cannot represent the latest active period of the fault. This put to us a significant question in regard to the age determination and study of active faults. Other noticeable questions are also inferred to in this study.

Study on relationship between deep and shallow structures along north boundary fault of Yanqing-Fanshan basin

Based on the results of surface geology, shallow and deep seismic survey, features of micro-earthquake activity along the north boundary fault of Yanqing-Fanshan sub-basin and their relationship with the surface active faults and the deep-seated crustal structure are analyzed using the recordings from the high-resolution digital seismic network. The focal mechanism solutions of micro-earthquakes, whose locations are precisely determined by the seismic network, have confirmed the structural characteristics to be the rotational planar normal fault and demon-strated the surface traces of the north boundary fault of Yanqing-Fanshan sub-basin. By using the digital recordings of earthquakes with the high resolutions and analyzing the mechanism solutions, our study has revealed the rela-tionship between the geological phenomena in the shallow and deep structures in Yanqing-Huailai basin and the transition features from the brittle to ductile deformation with the crustal depth.

Research on the Characteristics of Deep and Shallow Structures in Shenyang City and Their Relationship with Seismic Activities

Deep and shallow tectonic data in Shenyang and its relationship with seismic activity shows that the NE trending faults developed on the surface control the formation and development of the fault-uplift and fault-depression.The uplift and depression of the bedrock at a depth of 7km underground are consistent with the surface structure.12 planar listric normal faults have developed above a depth of 18km ～ 20km and two deep faults have developed in the lower crust.Because of the deep incision and new activities,the surface Wanggangpu-Xinchengzi fault and Yongle-Qingshuitai fault,which correspond to the deep F3 fault and F6 fault,might be related to seismic activity in Shenyang.

P-wave tomography and relation between shallow and deep structures beneath the Songliao basin

We selected relative travel-time residuals from teleseismic waveform data using the waveform correction method and imaged the P wave velocity structure beneath Northeast China. In combination with other geophysical data, we discussed the relation between the shallow and deep structures of the area. The results show that there is a primary high-velocity zone with some high- and low-ve- locity distribution characters beneath the Songliao basin. The low-velocity anomalies may extend down to the upper mantle, and may be the result of material upwelling. The low-velocity anomaly beneath the southern part of the Songliao basin is connected to those beneath the Chang- baishan and A＇ershan volcanic areas. It may be an upwelling channel from the mantle beneath the Songliao basin and adjacent area. This finding indicates the Songliao basin was a result of asthenospheric upwelling caused by subduction of the Pacific plate under the Eurasian plate.

Hydrodynamic Links between Shallow and Deep Mineralization Systems and Implications for Deep Mineral Exploration

Deep mineral exploration is increasingly important for finding new mineral resources but there are many uncertainties.Understanding the factors controlling the localization of mineralization at depth can reduce the risk in deep mineral exploration.One of the relatively poorly constrained but important factors is the hydrodynamics of mineralization.This paper reviews the principles of hydrodynamics of mineralization,especially the nature of relationships between mineralization and structures,and their applications to various types of mineralization systems in the context of hydrodynamic linkage between shallow and deep parts of the systems.Three categories of mineralization systems were examined,i.e.,magmatic-hydrothermal systems,structurally controlled hydrothermal systems with uncertain fluid sources,and hydrothermal systems associated with sedimentary basins.The implications for deep mineral exploration,including potentials for new mineral resources at depth,favorable locations for mineralization,as well as uncertainties,are discussed.

The heterogeneous characteristics of crust-mantle structures and the seismic activities in the northwest Beijing region

In this paper, the abnormal characteristics of the crustal structures in the seismic active region, Yanqing-Huailai and Zhangbei-Shangyi, are obtained by means of comprehensively interpreting and studying the data of deep seis- mic sounding profiles passing through the northwestern part of Zhangjiakou-Bohai seismic zone. The results show that the fluctuation of crystalline basement in the study region is obvious and that there exist considerable differ- ences in depth in different geological units. The locally abrupt variation of crystalline basement depths may be regarded as a mark of existence of crystalline basement faults. These crystalline basement faults and deep crustal faults provide a pass for the magma upwelling, resulting in the strong inhomogeneity of crustal structures. These phenomena of the complex seismic reflected waves and locally discontinuous reflection zones with different en- ergy indicate that the intensive squeeze and deformation of crust took place, which have led to the complex crustal structures and offered the dynamic source for the earthquake occurrence in this region. The low velocity bodies in different depths of crust and the local interface C1 in Zhangbei-Shangyi region may result from repeated magmatic activities. The certain stress accumulation in the brittle upper crust can cause the occurrence of earthquake under the action of local tectonic activity.

SpectraTr:A novel deep learning model for qualitative analysis of drug spectroscopy based on transformer structure

The drug supervision methods based on near-infrared spectroscopy analysis are heavily dependent on the chemometrics model which characterizes the relationship between spectral data and drug categories.The preliminary application of convolution neural network in spectral analysis demonstrates excellent end-to-end prediction ability,but it is sensitive to the hyper-parameters of the network.The transformer is a deep-learning model based on self-attention mechanism that compares convolutional neural networks(CNNs)in predictive performance and has an easy-todesign model structure.Hence,a novel calibration model named SpectraTr,based on the transformer structure,is proposed and used for the qualitative analysis of drug spectrum.The experimental results of seven classes of drug and 18 classes of drug show that the proposed SpectraTr model can automatically extract features from a huge number of spectra,is not dependent on pre-processing algorithms,and is insensitive to model hyperparameters.When the ratio of the training set to test set is 8:2,the prediction accuracy of the SpectraTr model reaches 100%and 99.52%,respectively,which outperforms PLS DA,SVM,SAE,and CNN.The model is also tested on a public drug data set,and achieved classification accuracy of 96.97%without preprocessing algorithm,which is 34.85%,28.28%,5.05%,and 2.73%higher than PLS DA,SVM,SAE,and CNN,respectively.The research shows that the SpectraTr model performs exceptionally well in spectral analysis and is expected to be a novel deep calibration model after Autoencoder networks(AEs)and CNN.

Seismotectonic Study on the West Part of the Interaction Zone Between Southern Tianshan and Northern Tarim

The interaction zone between southern Tianshan and northern Tarim is located at the northeast side of Pamir. It is a region with high seismicity. We constructed a seismotectonic model for the west part of this zone from geological profiles, deep crust seismic detection and earthquake focal mechanisms data. Based on the synthesized geological features, deep crust structure, and earthquake focal mechanisms, we think that the main regional tectonic feature is that the Tianshan tecto-lithostratigraphic unit overthrusts on the Tarim block. The Tianshan tectonic system includes the Maidan fault and thrust sheets in front of the fault; The Tarim tectonic system includes the underground northern Tarim margin fault, conjugate faults in basement and overthrust fault in shallow. The northern Tarim margin fault is a high angle fault deep in the Tarim crust, adjusting different trending deformation between Tianshan and Tarim. It is a major active fault that can generate large earthquakes. The other faults, such as the Tianshan overthrust system and the Tarim basement faults in this area may generate moderately strong earthquakes with different styles.

深、浅部煤层气地质条件差异性及其形成机制

深部煤层气资源丰富、开发前景广阔,但对其与浅部煤层气地质条件的内在联系研究尚不够深入。从煤层形成演化角度出发,以鄂尔多斯盆地上古生界煤层为例,总结了煤层深埋深藏型、深埋浅藏型及浅埋浅藏型3种埋深演化模式。系统分析了深部和浅部煤层在温压条件与含气性、地应力与渗透率特征、变质程度与含水性等方面的差异性及其形成机制。研究指出,受埋深与演化过程影响,深部和浅部煤储层温度最多相差100℃以上,储层压力最大相差40 MPa左右,导致由浅部向深部,气体赋存状态以吸附气为主转变为吸附气与游离气共存,地应力场由水平应力主导转化为垂向应力主导,煤储层孔隙率、渗透率及含水性逐渐降低。明确了深部煤层气的典型特点,即:在高温高压条件下,以吸附态和游离态共存于一定深度以下煤储层中的甲烷气体,该类煤储层在垂向应力为主导的作用下,孔裂隙空间极度压缩,含水极少且矿化度极高,内生微裂隙为主要渗流通道。基于含气性临界深度和地应力场转换深度的不一致性,指出浅部向深部煤层演化过程中存在过渡区,该区内呈现出非典型深部煤层气的特点,或深部煤层气和浅部煤层气地质条件共存的情况,在勘探开发过程中,应具体分析,制定针对性开发方案,以实现浅部与深部煤层气的高效协同开发。

基于改进Faster R-CNN的变电站设备外部缺陷检测

针对变电站设备外部缺陷目标检测任务中目标形状多样,周围环境复杂,当前代表性算法识别准确度低,错检漏检严重的问题,对比了众多目标检测算法在变电站设备缺陷数据集上的检测结果,检测精度较高的是添加了特征融合金字塔结构的Faster R-CNN(faster region-based convolutional network)算法,但其对小目标物体和设备渗漏油的检测精度仍有提升空间,为此设计一种基于Faster R-CNN的改进算法。改进算法通过对输入图像进行数据增强,在网络中添加SPP(spatial pyramid pooling)结构以及改进特征融合方式,对分类以及边界框回归损失函数进行改进的方式来提高缺陷的检测精度。与原Faster R-CNN算法进行对比,改进算法在变电站设备缺陷目标检测数据集的检测结果中AP(average precision)(0.5∶0.95)提高了2.7个百分点,AP(0.5)提高了4.3个百分点,对小目标物体的检测精度也提高了1.8个百分点,试验结果验证了该方法的有效性。

面向图像分类的视觉Transformer研究进展

Transformer是一种基于自注意力机制的深度学习模型,在计算机视觉中展现出巨大的潜力。而在图像分类任务中,关键的挑战是高效而准确地捕捉输入图片的局部和全局特征。传统方法使用卷积神经网络的底层提取其局部特征,并通过卷积层堆叠扩大感受野以获取图像的全局特征。但这种策略在相对短的距离内聚合信息,难以建立长期依赖关系。相比之下,Transformer的自注意力机制通过直接比较特征在所有空间位置上的相关性,捕捉了局部和全局的长距离依赖关系,具备更强的全局建模能力。因此,深入探讨Transformer在图像分类任务中的问题是非常有必要的。首先以Vision Transformer为例,详细介绍了Transformer的核心原理和架构。然后以图像分类任务为切入点,围绕与视觉Transformer研究中的性能提升、计算成本和训练优化相关的三个重要方面,总结了视觉Transformer研究中的关键问题和最新进展。此外,总结了Transformer在医学图像、遥感图像和农业图像等多个特定领域的应用情况。这些领域中的应用展示了Transformer的多功能性和通用性。最后,通过综合分析视觉Transformer在图像分类方面的研究进展,对视觉Transformer的未来发展方向进行了展望。

利用图像平滑结构信息指导图像修复

利用图像结构特征进行图像修复,是近年来在深度学习技术广泛应用背景下出现的新方法。应用该方法可以在缺失区域内生成合理的内容,但图像修复结果过于依赖图像结构的提取内容,且在实际训练中会出现错误的持续传播和累积,一旦图像结构存在噪声或失真会直接影响到图像的生成质量。该方法处在探索应用阶段,尚存在网络训练难度大、鲁棒性较差、生成图像上下文语义不一致等问题。为此,本文提出了一种图像平滑结构指导修复的并行网络结构。图像平滑结构的生成内容不直接作为下一级网络的输入,只为网络的解码层提供指导信息。同时,为了更好地匹配和均衡结构与图像之间的特征关系,本文结合transformer提出了一种多尺度特征指导模块。该模块利用transformer联系全局特征的强大建模能力,对结构和图像纹理之间的特征进行匹配和均衡。实验结果表明,本文方法在三个常用的数据集上能够有效地恢复图像缺损内容,并且可以作为图像编辑工具实现目标移除。

受邻近开挖影响的浅基础砌体结构三阶段分析法

浅基础砌体结构的材料力学行为和受力体系决定了其对变形极其敏感,尤其是邻近深基坑工程施工时,需在基坑开挖前对影响范围内的砌体结构建筑进行安全评价。鉴于此,提出一种用于受邻近深开挖影响的浅基础砌体结构变形和损伤分析的三阶段分析法。三阶段分析法依托于数值模拟,根据影响因子权重依次构建地基结构模型、荷载结构模型和分离式砌体模型,在减少计算量的同时,尽可能准确预测砌体结构的变形和损伤风险。本文依托杭州市某深基坑工程,采用该分析法对受深开挖影响的砌体结构建筑进行深入分析,结合实测数据验证计算结果。结果表明,三阶段分析法能较好地反映因基坑开挖引起的周边建筑物的变形和表观损伤情况,在评价浅基础砌体结构损伤、预测结构变形等方面有较好的适用性,可为基坑开挖与周边建筑物相互影响的超前预测提供一种思路。

多时态BIT遥感图像建筑物的变化检测

针对来自相同地理空间的不同时刻遥感图像之间的季节性和光度变化(色差)等因素所引起的干扰,提出了多时态-BIT遥感图像变化检测方法.该方法引入了过去多个不同时刻的遥感图像,融合当前遥感图像与过去时态遥感图像两两变化检测的结果,该方法有助于排除季节性和光度变化引起的误报,提高了变化检测的准确性;并且利用过去多个不同时刻的遥感图像,进一步消除非目标建筑变化的影响,其变化点像素差值引入作为损失函数正则化项,从而进一步提高变化检测的鲁棒性和可靠性.本文以三时态(3个不同时刻的遥感图像)为例,使用了遥感图像建筑物变化数据集进行了实验.实验结果表明,多时态-BIT方法相对于仅考虑两个时态的变化检测方法,在遥感图像建筑物变化检测任务中表现出更好的效果.

背景噪声频率-贝塞尔变换法研究厦门岛浅层S波速度结构

为获取厦门岛浅层三维速度结构,本文使用福建省地震局于2019年布设的50个流动地震台站组成的密集台阵所记录到的垂向分量波形数据,计算背景噪声互相关,利用频率-贝塞尔变换(F-J)法提取子台阵的瑞利波相速度频散曲线进行反演,获取子台阵下方一维S波速度结构,并通过插值得到厦门岛4 km以内的三维S波速度模型.结果表明,整体上厦门岛S波速度随深度增加呈现递增趋势,在近地表处的横向不均匀性显著.0~1 km深度的S波速度大体上呈现“北高南低”的特征,与岩性分布有较好的对应关系.随着深度的递增,高低速特征分布逐渐呈现为连续性较好,近北东向展布的条带状.结合前人研究结果,本文推测这一现象与厦门岛岩层在不同时期受到挤压和拉张作用形成的褶皱和断裂相关,反映了厦门岛所经历的多次构造变动和岩浆侵入事件.垂向剖面的速度差明显区与研究区断裂带位置具有一定的对应关系.研究结果可为厦门岛的构造演化研究和地质灾害评估提供新的科学依据.

结合金字塔结构和注意力机制的单目深度估计

单目深度估计是由单幅彩色图像预测出一幅稠密的深度图像。针对目前单目深度估计算法存在边界模糊、上下文信息捕捉能力不足等问题,提出了一种结合金字塔结构和注意力机制的单目深度估计算法。算法采用编码器-解码器的总体框架,其中编码器选用PVTv2网络,目的是利用Transformer网络在建模全局信息方面的优势以获取更充分的全局语义信息;解码器由深度估计主分支和2个金字塔子分支组成。深度估计主分支通过空间和通道注意力机制来自适应地关注编码器和解码器特征间重要的特征区域和特征通道;拉普拉斯金字塔子分支和深度残差金字塔子分支旨在从彩色图像和深度估计主分支深度特征中学习到丰富的局部信息并传递到深度估计主分支,进一步解决单目深度估计中细节缺失、结构混乱等问题。实验结果表明,与先进的算法P3Depth相比,在室内公开数据集NYU Depth V2上,该算法的δ_(1.25)阈值精度提升了1.22%,绝对误差和根均方误差分别降低了5.8%和2.8%;而在室外公开数据集KITTI上,该算法的绝对误差、根均方对数误差和根均方误差分别降低了8.5%,3.9%和0.4%。该算法提升了深度估计精度并得到了良好的视觉呈现效果。

基于Transformer的视觉目标跟踪方法综述

视觉目标跟踪是计算机视觉中的重要任务之一,为实现高性能的目标跟踪,近年来提出了大量的目标跟踪方法,其中基于Transformer的目标跟踪方法由于具有全局建模和联系上下文的能力,是目前视觉目标跟踪领域研究的热点。首先,根据网络结构的不同对基于Transformer的视觉目标跟踪方法进行分类,概述相关原理和模型改进的关键技术,总结不同网络结构的优缺点;其次,对这类方法在公开数据集上的实验结果进行对比,分析网络结构对性能的影响,其中MixViT-L(ConvMAE)在LaSOT和TrackingNet上跟踪成功率分别达到了73.3%和86.1%,说明基于纯Transformer两段式架构的目标跟踪方法具有更优的性能和更广的发展前景;最后,对方法当前存在的网络结构复杂、参数量大、训练要求高和边缘设备使用难度大等不足进行总结,并对今后的研究重点进行展望,通过与模型压缩、自监督学习以及Transformer可解释性分析相结合,可为基于Transformer的视觉目标跟踪提出更多可行的解决方案。

双侧基坑紧邻地铁车站施工安全影响和保护对策

以南京宁句城际线某车站两侧紧邻的3#、4#基坑为例,运用MADIS/GTS三维有限元软件分析模型研究3#(深)、4#(浅)基坑双侧不同施工时序下地铁结构位移变化规律。研究表明:地铁结构主要发生面向基坑一侧的水平位移;3#基坑主要影响车站主体东侧和1号风亭;4#基坑主要影响车站主体西侧和1号出入口。综合来说,先施工3#深基坑后施工4#浅基坑相比先施工4#浅基坑后施工3#深基坑对地铁车站、附属(风亭、出入口)产生的水平位移大,3#、4#基坑同时施工对地铁车站、附属影响介于3#、4#先后施工之间。