Change in Grain-Size Composition of Lignite under Cyclic Freezing-Thawing and Wetting-Drying

The paper presents the change in grain-size composition of lignite under cyclic freezing-thawing (FTC) and wetting-drying (WDC). The article shows that in the spring and autumn periods the lignites can be subjected to repeated freezing-thawing and wetting-drying, which determines the possibility of changing their grain-size composition and structure. Experimental studies in laboratory conditions on the influence of cyclic freezing-thawing (FTC) and wetting-drying (WDC) on the quality indicators of lignites have been carried out, their granulometric (fractional) composition has been studied. Freezing-thawing cycle conditions are as follows (FTC): minimum exposure temperature: -20°C;maximum: +5°C;relative humidity: 30%;number of processing cycles: 3. Wetting-drying cycles are as follows (WDC): drying temperatures are +20, +40, +60, +80°C, drying time 90 minutes, the coals are further subjected to rain (soaking) for a period of water saturation to humidity of 30% - 40% and dry again. The number of wetting-drying cycles is 3 times. The tests have revealed the destructive effects of FTC and WDC on the samples of lower metamorphic grade coal, and the cycles of wet-dry lead to the much higher yield of fine sizes (-6+0;-13+0 mm) than the cycles of freeze-thaw. Furthermore, it is found that the increase in the yield of fines depends on the heating temperature: coal disintegration proceeds more intensively at a higher temperature of drying.

Balance Analysis of Profit and Loss on Exploiting Coal-Bed Gas in China

Models about four aspects according to the balance principle and practice in China were established, which involve the minimum production scale, alert production scale, safe running scale, and the goal production scale for specified profit level. It provides an effective quantitative analyzing method for the investors of coal bed gas exploitation project.

How moisture loss affects coal porosity and permeability during gas recovery in wet reservoirs?

Moisture in coal seams changes gas adsorption capacity, induces coal deformation, and affects coal porosity. However, fewer studies have investigated the dynamic process of moisture loss. In this study, a fully coupled multi-physical model for coal deformation, gas flow and moisture loss was implemented. It validated the coal-gas-moisture interactions of the decay of gas adsorption capacity and the coal shrinkage.Subsequently, the proposed model was applied to a simulation of coal seam gas recovery from wet reservoir and solved using the finite method in COMSOL Multiphysics 3.5. Analyses of the component factors and the sensitive parameters of moisture loss on coal porosity and permeability were comprehensively studied at last. The results reveal that moisture loss enhances coal porosity and permeability. The decay of gas adsorption capacity decreases coal permeability while the coal shrinkage promotes it. The decrease of the adsorption decay coefficient and the increase of the initial density of saturated water vapor and water evaporation constant can enhance the permeability of wet coal seams.

Bacterial degradation of coal discard and geologically weathered coal

The biodegradation of coal discard is being intensively studied in South Africa in an effort to develop passive methods for the successful revegetation and rehabilitation of waste dumps, to mitigate pollution, and facilitate mine closure. Bacteria were isolated from slurries of coal tailings and diesel-contaminated soil, screened for coal biodegradation competence, characterized, and the colonization and degradation of coal discard and geologically weathered coal investigated using individual isolates and consortia. Ten novel coal-degrading bacterial strains were isolated and characterized, the gene sequences deposited with GenBank, and the (wild-type) strains deposited at Microbial Culture Collection, India. The results from the present work show that bituminous coal discard and geologically weathered coal is used by these isolates as carbon and energy source. Isolated strains and consortia colonized and degraded both coal substrates. Growth rate of the isolates is faster and stationery phase achieved sooner in minimal medium containing geologically weathered coal. This observation suggests that the oxygen-rich weathered coal is a more friable substrate and thus readily colonised and biodegraded. A reduction in mass of substrate is demonstrated for both individual isolates and consortia. The changes in pH and associated media colouration occurred concomitant with formation of humic acid-like (HS) and fulvic acid-like substances (FS) which is confirmed following analysis of these products by FT-IR spectroscopy. It is concluded that preferential metabolism of alkanes from the coal substrates provided the carbon and energy for bacterial growth and transformation of the substrates to HS and FS.

Experimental evaluation of activated carbon derived from South Africa discard coal for natural gas storage

Lacking in literature is the use of discard coal to produce activated carbon and in its subsequent use in the storage of natural gas. In this study, the characterization and gas storage evaluation of a largely porous activated carbon with large surface area synthesized from discard coal were investigated. Discard coals are waste material generated from coal beneficiation process. In developing the activated carbon, chemical activation route with the use of KOH reagent was applied. The effects of KOH/discard coal weight ratio (1:1, 2.5:1, 4:1), temperature (400-800 ℃) and particle size (0.15-0.25 mm, 0.25-0.5 mm, 0.5-1 mm) on the adsorptive properties of the activated carbon were methodically evaluated and optimized using response surface methodology. The synthesized activated carbon was characterized using BET, SEM/EDS, and XRD. The results showed that for each activation process, the surface area and pore volume of the resulting activated carbon increased with increased temperature and KOH/discard coal weight ratio. The maximum surface area of 1826.41 m2/g, pore volume of 1.252 cm^3/g and pore size of 2.77 nm were obtained at carbonization temperature of 800 ℃ and KOH/discard coal weight ratio of 4:1. Methane and nitrogen adsorption data at high pressure were fitted to Toth isotherm model with a predictive accuracy of about 99%. Adsorption parameters using the Toth model provides useful information in the design of adsorbed natural gas storage system. According to the requirements of adsorbent desired for natural gas storage, it could be stated that the synthesized activated carbon could well be applied for natural gas storage.

Fractal characteristics of electric properties of coal

In the light of fractal geometry theory, the characteristics of coal's electric parameters (including dielectric constant, alternating conductivity, dielectric loss angle tangent and electric polarization constant) were studied by using literature data. The results are shown that the electrical properties of coal have fractal characteristic. The fractal dimensions of dielectric, alternating conductivity, dielectric loss angle tangent were obtained, and are relative to the content of pyrite sulfur, heat and ash content of coal.

基于CED-YOLOv5s模型的煤矸识别方法研究

由于煤矿井下高噪声、低照度、运动模糊的复杂工况和煤矸易聚集现象,导致煤矸目标检测模型特征提取困难及煤矸分类、定位不准确问题。针对该问题,提出一种基于CED-YOLOv5s模型的煤矸识别方法。首先,在YOLOv5s主干网络中引入坐标注意力(CA)机制,通过将坐标信息嵌入信道关系和长程依赖关系中对特征图进行编码,充分利用通道注意力信息和空间注意力信息,使模型更加关注重要特征,抑制无用信息。其次,在YOLOv5s的检测头部引入EIoU回归损失函数,将目标框与锚框的宽高差异最小化,以增强目标的位置和边界信息,提高模型在密集目标下的定位精度和收敛速度;最后,在YOLOv5s的检测头部引入轻量化解耦头,解耦出单独的特征通道,分别用于分类任务和回归任务,解决了原模型中耦合头部分类任务与回归任务的相互干扰问题,进一步提升了模型的并行运算效率与检测精度。实验结果表明:CED-YOLOv5s模型与其他YOLO系列目标检测模型相比,综合性能最佳,平均检测精度达94.8%,相较于YOLOv5s模型提升了3.1%,检测速度达84.8帧/s,可充分满足煤矿井下煤矸实时检测需求。

基于CFS-YOLO算法的复杂工况环境下煤矸图像识别方法

针对煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂工况环境因素导致煤矸识别存在误检、漏检以及检测精度低的问题,提出一种基于CFS-YOLO算法的煤矸智能识别模型。采用ConvNeXt V2(Convolutional Neural Network with NeXt Units Version 2)特征提取模块替换主干网络末端的2个C3(Cross Stage Partial Bottle Neck Mudule)模块,通过将掩码自动编码器(Masked Autoencoders,MAE)和全局响应归一化(Global Response Normalization,GRN)层添加到ConvNeXt架构中,有效缓解特征崩溃问题以及保持特征在网络传递过程中的多样性;采用Focal-EIOU(Focal and Efficient Intersection Over Union)损失函数替换原CIOU(Computer Intersection Over Union)损失函数,通过其Focal-Loss机制和调整样本权重的方式优化边界框回归任务中的样本不平衡问题,提高模型的收敛速度和定位精度;添加无参注意力机制(Simple Attention Mechanism,SimAM)于主干网络每个C3模块的后端,凭借其注意力权重自适应调整策略,提升模型对尺度变化较大或低分辨率煤矸目标关键特征的提取能力。通过消融试验和对比试验验证所提CFS-YOLO模型的有效性与优越性。试验结果表明:CFS-YOLO模型对于煤矸在煤矿高噪声、低照度、运动模糊与大批量煤矸混杂等复杂环境下的检测效果均得到有效提高,模型的平均精度均值达到90.2%,相较于原YOLOv5s模型的平均精度均值提高了3.7%,平均检测速度达到90.09 FPS,可充分满足煤矸实时检测的需求。同时与YOLOv5s、YOLOv7-tiny与YOLOv8n等6种YOLO系列算法相比,CFS-YOLO模型对煤矿复杂环境的适应性最强且综合检测性能最佳,可为煤矸的智能高效分选提供技术支持。

可变粘压裂液研究及在深煤层中应用

针对深煤层压裂开发需要,研发一套可变粘压裂液体系,通过粘度、降阻率、表面张力、界面张力等指标优选构建了低粘-中粘-高粘滑溜水体系,并对体系滤失性能、破胶性能和岩心伤害性能进行了评价。实验结果表明:随着体系粘度增加,降阻率先增加后降低;高粘度有助于降低滤失。通过调节破胶剂用量实现破胶可控,对煤岩具有较小的岩心伤害率。配合体积改造压裂工艺,可有效助力深煤层开发。优选的可变粘压裂液具有较好的推广和应用前景。

基于YOLOv5s−FSW模型的选煤厂煤矸检测研究

针对现有煤矸检测模型存在的特征提取不充分、参数量大、检测精度低且实时性差等问题,提出了一种基于YOLOv5s−FSW模型的选煤厂煤矸检测方法。该模型在YOLOv5s的基础上进行改进,首先将主干网络的C3模块替换为FasterNet Block结构,通过降低模型的参数量和计算量提高检测速度;然后,在颈部网络引入无参型SimAM注意力机制,增强模型对复杂环境下重要目标的关注,进一步提高模型的特征提取能力;最后,在输出端用Wise−IoU替换CIoU边界框损失函数,使模型聚焦普通质量锚框,提高收敛速度和边框的检测精度。消融实验结果表明:与YOLOv5s模型相比,YOLOv5s−FSW模型的平均精度均值(mAP)提高了1.9%,模型权重减少了0.6 MiB,参数量减少了4.7%,检测速度提高了19.3%。对比实验结果表明:YOLOv5s−FSW模型的mAP达95.8%,较YOLOv5s−CBC,YOLOv5s−ASA,YOLOv5s−SDE模型分别提高了1.1%,1.5%和1.2%,较YOLOv5m,YOLOv6s模型分别提高了0.3%,0.6%;检测速度达36.4帧/s,较YOLOv5s−CBC,YOLOv5s−ASA模型分别提高了28.2%和20.5%,较YOLOv5m,YOLOv6s,YOLOv7模型分别提高了16.3%,15.2%,45.0%。热力图可视化实验结果表明:YOLOv5s−FSW模型对煤矸目标特征区域更加敏感且关注度更高。检测实验结果表明:在环境昏暗、图像模糊、目标相互遮挡的复杂场景下,YOLOv5s−FSW模型对煤矸目标检测的置信度得分高于YOLOv5s模型,且有效避免了误检和漏检现象的发生。

典型炼焦煤成焦界面结合特性及其焦炭高温强度的研究

焦炭的界面结合对提高焦炭质量有十分重要的作用。实验选用五种炼焦煤分别配入弱粘结性的XS贫瘦煤进行坩埚焦实验,随着煤料粒度增大和贫瘦煤含量增加,炼焦煤界面结合指数下降。肥煤自身黏结性好,界面结合指数影响较小,而1/3焦煤和焦煤则影响较大。同时分别完成了LF煤和XQ煤与XS贫瘦煤的共炭化实验,表明随着界面结合指数的提高,焦炭的高温热性能得到了较好改善。相应地,微晶结构更加规整,碳层堆积高度和碳网平面直径增大。

基于机器视觉的煤矸石分选方法研究

传统的煤矸石分选方法存在成本高、效率低以及安全性不足等问题。近年来,随着深度学习相关技术的迅速发展,基于目标检测算法的智能选矸已经成为矸石分选的重要研究方向。为实现矸石与煤块的高效分选,本文提出一种基于机器视觉与深度学习相结合的检测方法。该方法以YOLOv5s模型为基础,首先,在主干部分中加入卷积注意力模块(CBAM)用于提高网络的特征提取能力;其次,在颈部网络部分采用加权双向特征金字塔结构(BiFPN)来增强网络的多尺度特征融合,避免漏检与误检现象的发生;再次,在预测部分使用EIoU函数作为改进后模型的损失函数,以进一步提高检测精度;最后,在训练前对原有数据集进行扩充,使模型的泛化能力得到进一步加强。实验结果表明:改进后模型平均检测精度为95.3%,较原模型提高了2.1%,能够有效地替代人工分选。

基于改进YOLO v8的煤中杂物检测研究

针对选煤厂毛煤中夹杂的铁丝、编织袋、木头、网片等杂物会对智能干选设备运行及后续生产环节造成严重影响的问题,提出一种基于改进YOLO v8的手选胶带杂物识别方法。引入全局注意力机制,增强图像跨维度特征交互;引入加权双向特征金字塔网络结构,通过自适应控制不同尺度特征图之间的融合,提高模型对杂物的多尺度检测能力。在此基础上,采用WIoU损失函数替换CIoU损失函数,改善模型训练过程中样本质量的平衡问题,以提高模型的性能。通过数据增强扩充煤中杂物数据集,依据实验验证改进YOLO v8的结果。实验结果表明,改进后的算法与原YOLO v8相比,对手选胶带煤中杂物的平均检测精度明显提高,为毛煤入选前的预先智能除杂奠定了基础。

热辐射强度对煤燃烧特性的影响实验研究

为研究温度对煤燃烧特性的影响,采用锥形量热仪在不同热辐射强度(35、40、45、50 kW/m^(2))下对朱集东煤矿气煤进行燃烧测试。结果表明:煤样热释放速率峰值随热辐射强度增加而增大,热辐射强度50 kW/m^(2)时煤样热释放速率峰值为81.54 kW/m^(2),较35、40和45 kW/m^(2)时分别增加15.96%、10.41%和1.53%;煤样总热释放量变化顺序为45 kW/m^(2)组>50 kW/m^(2)组>40 kW/m^(2)组>35 kW/m^(2)组,即随热辐射强度增加先增大后减小。随着热辐射强度由35 kW/m^(2)增加至50 kW/m^(2),煤样质量损失速率峰值与总质量损失量均呈上升趋势,分别增加了38.15%和13.75%。在燃烧后期,45 kW/m^(2)组的CO释放量峰值最小,为0.36 kg/kg,相较于35、40和50 kW/m^(2)组分别降低了0.33、0.27和0.04 kg/kg。

煤矿采空区覆岩“三带”智能识别方法

为了探明煤矿采空区覆岩破坏类型与裂隙发育情况,快速准确划分采空区覆岩弯曲带、断裂带、垮落带(简称“三带”),提出智能识别方法,为制定采空区治理方案提供依据。以山东某矿区采空区为例,采用贝叶斯在线变化点检测(Bayesian Online Changepoint Detection,BOCD)算法对钻进过程中的冲洗液漏失量和钻速数据在煤矿采空区“三带”界限处的变化及响应特征进行分析。以勘察规范中有关“三带”高度计算经验公式作为约束条件,对冲洗液漏失量和钻速数据中的候选变化点进行检测、优选,进而确定煤矿采空区“三带”界限深度。智能识别的结果与实际值吻合,其中,弯曲带下限、断裂带下限、垮落带下限的深度误差分别为+0.67、+0.31和+0.52m,弯曲带、断裂带和垮落带的高度误差分别为+0.14%、-0.63%和+2.49%。基于钻进数据的采空区覆岩“三带”智能识别方法精度满足设计需要,切实可行。该方法将钻进数据与经验公式相结合,在钻进过程中即可完成“三带”界限的划分,充分发挥数据的时效性,避免了技术人员主观判断对识别结果产生影响。相较于依靠多种方式综合确定“三带”界限的传统方法,该智能识别方法显著提高了“三带”划分的时效性和准确性。

基于YOLO- RepGFPN模型的煤矸检测方法研究

现有基于深度学习的煤矸识别方法在光线昏暗、高噪声、遮挡等因素的干扰下容易出现漏检、错检的现象。针对该问题,提出一种基于YOLO-RepGFPN模型的煤矸检测方法。YOLO-RepGFPN模型基于YOLOV5s模型改进:1)在Backbone部分中引入CA注意力机制,使网络能充分利用丰富的上下文信息,增强特征学习能力,2)在Neck部分采用RepGFPN网络模块,增强模型对煤矸高级语义和低级空间特征的提取和融合能力,从而提高检测精度,3)修改损失函数为XIOU增强目标匹配和目标边框回归过程中的定位精度和鲁棒性。实验结果表明:RepGFPN煤矸识别方法与目前主流YOLO算法相比识别精度(mAP)最高,达到94.7%,可有效避免恶劣条件下,煤矸识别时容易出现的漏检、误检和重检现象。

基于MES−YOLOv5s的综采工作面大块煤检测算法

综采工作面的目标具有高速运动、多尺度、遮挡等特点,现有的目标检测算法存在精度低、模型占用的内存大、硬件依赖强等问题。针对上述问题,提出了一种基于MES−YOLOv5s的综采工作面大块煤检测算法。采用轻量化设计,将MobileNetV3作为主干网络,以减小模型占用的内存,提高CPU端的检测速度;在颈部网络添加高效多尺度注意力(EMA)模块,融合不同尺度的上下文信息,并进一步减少计算开销;采用SIoU损失函数代替CIoU损失函数,以提高训练速度和推理准确性。消融实验结果表明:MobileNetV3大幅减少了模型占用的内存和检测时间,但mAP损失严重;EMA模块和SIoU损失函数可在一定程度上恢复损失的精度,同时保证模型在CPU上具有较高的检测速度,满足煤矿井下目标实时检测需求。对比实验结果表明,与DETR,YOLOv5n,YOLOv5s,YOLOv7模型相比,MES−YOLOv5s模型综合性能最好,mAP为84.6%,模型占用的内存为11.2 MiB,在CPU端的检测时间为31.8 ms,在高速运动、多尺度、遮挡和多目标的工况环境下能够保持较高的召回率和精度。

神华煤直接液化残渣理化特性与分子结构特征研究

煤液化残渣制备高端炭材料是目前煤直接液化技术领域研究的热点,该过程实现的关键在于煤液化残渣的深度脱灰,而深度脱灰工艺技术开发需要对液化残渣理化特性与分子结构有深入的认识。以神华煤液化残渣为研究对象,借助TG-DTG探究了液化残渣的失重特性;借助激光粒度仪和透射电镜(TEM)分别研究了液化残渣在洗油和四氢呋喃两种溶剂中的粒度分布特性和颗粒形貌特征;在正负离子(±ESI)和正离子大气压光电离(+APPI)模式下,利用傅里叶变换离子回旋共振质谱仪(FT-ICR MS)对液化残渣的分子结构组成进行了鉴别。研究结果表明:液化残渣的主要失重温区为300℃~600℃,466℃时失重速率最大,此时的活性最高;液化残渣在洗油或四氢呋喃中的粒度呈现百纳米级正态分布,残渣在洗油中粒度分布更为集中且粒径更小;液化残渣中含丰富的N1、N2类含氮杂环以及O1、O2、O3、CH类含氧多环的芳香分子构型。