基于GPR与ESR的煤岩性状识别方法研究.pdf
博士学位论文 基于 GPR 与 ESR 的煤岩性状识别方法研究 Study of Coal-Rock Characteristics Identification Based on GPR and ESR 作 者苗曙光 导 师刘晓文 教授 中国矿业大学 二○一九年六月 万方数据 学位论文使用授权声明 学位论文使用授权声明 本人完全了解中国矿业大学有关保留、使用学位论文的规定,同意本人所撰 写的学位论文的使用授权按照学校的管理规定处理 作为申请学位的条件之一, 学位论文著作权拥有者须授权所在学校拥有学位 论文的部分使用权,即①学校档案馆和图书馆有权保留学位论文的纸质版和电 子版,可以使用影印、缩印或扫描等复制手段保存和汇编学位论文;②为教学和 科研目的,学校档案馆和图书馆可以将公开的学位论文作为资料在档案馆、图书 馆等场所或在校园网上供校内师生阅读、浏览。另外,根据有关法规,同意中国 国家图书馆保存研究生学位论文。 (保密的学位论文在解密后适用本授权书) 。 作者签名 导师签名 年 月 日 年 月 日 万方数据 中图分类号 TD80 学校代码 10290 UDC 622.7 密 级 公开 中国矿业大学 博士学位论文 基于 GPR 与 ESR 的煤岩性状识别方法研究 Study of Coal-Rock Characteristics Identification Based on GPR and ESR 作 者 苗曙光 导 师 刘晓文 教授 申请学位 工学博士 培养单位 信息与控制工程学院 学科专业 检测技术与自动化装置 研究方向 矿山物联网 答辩委员会主席 沈连丰 评 阅 人 盲评 二○一九年六月 万方数据 致谢 致谢 感谢我的导师刘晓文教授对我生活和学习上的指导和帮助,从硕士到博士, 师从八载, 收获颇丰, 感触亦深。 刘老师渊博的专业知识、 真诚和蔼的待人态度、 严谨的治学准则和敬业精神为我树立了一个优秀科研和教育工作者的典范, 是我 学习的楷模,令我受益终身同时,感谢恩师丁恩杰教授给予我无私的指导和细 致的关怀。在生活上,丁老师待人真诚,心胸宽广,平易近人,时时刻刻为我们 着想,是严师更像慈父;在工作中,丁老师为我们提供了优渥的科研环境和丰富 的项目实践机会。在我的研究陷入进退两难,举步维艰,即将放弃的时候,丁老 师鼓励我重新梳理思路,坚定信念,使我慢慢地走出了挫败的泥潭,这段求学的 长征之路因有丁老师的陪伴,我倍感幸福和感动。在此,谨向我的两位恩师致以 最崇高的敬意 感谢张申教授、赵小虎教授、华钢教授、徐钊教授、尹洪胜教授、李雷达教 授、唐守锋教授、徐永刚副教授、胡青松副教授、刘亚峰老师、赵端老师、孟磊 老师、胡克想老师,刘鹏老师、李娟娟老师以及其他我未列及的老师给予的支持 和帮助,与你们的交流使我获益良多。 感谢我的工作单位淮北师范大学对我攻读博士学位的大力支持, 感谢学院领 导和同事们四年以来给予我的包容和帮助, 特别要感谢李素文院长和李淮江教授 在研究方法上的指导,让我深受启发。 感谢清华大学蒋敏教授给予我 ESR 样品制作,实验方案设计和论文修改上 的指导和帮助,感谢上海加美华科贸有限公司陈剑锋工程师对 Micro-ESR 仪器 的培训指导。 感谢同课题组的师兄王昕博士, 在井下现场实验和样品收集整理方面提供的 帮助;感谢同实验室的刘忠育、张雷、俞啸、邓婧婧、张凯、刘晓明,王满意等 博士和虞婧硕士的各种建议和有益讨论,使我的思路更加开阔。 感谢我们成立的 “矿大 CS 讨论小组” 的王刚和胡延军老师, 张然、 宋泊明、 孙中廷、程婷婷、邹翔宇等博士同学,与你们每周一次的讨论让我获得很多的灵 感、找到了科研的方法和钥匙。 感谢我的父母,感谢你们给了我生命,抚育我长大成人,为我构筑舒适温暖 的家。从呱呱坠地到咿呀学语,从入学升学到择业择偶,父母无私的关爱陪伴了 我生命的每一个阶段。正是你们的陪伴,才让我一步一步的成长成才 感谢岳父岳母七年以来对我的关爱,谢谢你们帮我照顾孩子和家庭,让我能 够无后顾之忧地完成博士学业。 特别要感谢我的爱妻马玉红女士。 多年以来, 正是由于你担起了家庭的重担, 默默无闻地付出,我才有一个可爱聪明的女儿,一个温馨幸福的三口之家,让我 失败时不会感到孤单,迷茫时不会觉得无助。在此,我想发自心底地说一句“老 婆,你辛苦了” 在今后的岁月里,我一定会加倍地承担起家的责任。 感谢一些给予过我无私帮助,一些不知名的朋友,我们因网络而结缘,得到 了你们的支持和帮助,在此轻轻地道一声谢谢 衷心感谢各位专家、学者在百忙之中对我的论文进行审阅及指导,期待你们 的批评与指正 万方数据 I 摘 要 摘 要 我国煤炭工业发展的愿景是实现矿山的无人化和智能化,这也是国家能源 科技“十二五”规划(2011-2015) 确定的重点任务和目标,其要解决的关键问 题就是煤岩性状识别。煤矿安全生产“十三五”规划(2016-2020)提出推进煤矿 机械化、自动化、信息化、智能化改造,并逐步实现煤矿井下无人工作面,无人 值守;另外,国家重点基础研究计划(973 计划) (2014-2018) 深部危险煤层无 人采煤关键基础研究课题,将“煤岩性状在线识别与采掘状态感知原理及实现 方法”列为关键子课题,提出煤岩性状的在线识别是实现深部煤层无人采掘装备 自主控制最为关键的信息基础,是实现采煤机滚筒自动调高的依据。因此,研究 煤岩界面识别对于实现综采工作面智能化和无人化采煤具有重要意义。 煤岩识别 一直是一个研究热点,虽然国内外许多学者已经提出了大量的解决方案,但是这 些方法基本都停留在方法的可行性探讨阶段,缺乏对煤岩性状特征的研究,还没 有形成一个有效的解决方案。本文以煤岩界面和煤岩介质为研究对象,以煤岩介 质的电磁特性和 ESR (Electron Spin Resonance)自由基特性参数差异规律为研究 内容,利用电磁波技术开展了煤岩性状识别的相关研究,并完成了以下工作。 (1)以煤岩界面为研究对象,探索了基于电磁波技术的煤岩界面识别方法 及适用性问题,根据煤岩地质赋存条件特征和电性参数,提出了具有电磁参数的 煤岩界面模型并进行了理论分析研究。首先,根据煤岩界面模型,考虑到煤岩电 性差异和电磁波传播的衰减特性, 提出该模型的 GPR (Ground Penetrating Radar) 回波方程。 不但为 GPR 探测煤岩界面提供理论支撑, 而且为天线选型提供依据; 开展了不同煤岩组合, 不同 GPR 天线频率条件下的电磁探测适用性研究。 其次, 考虑到煤矿地质的复杂性,利用有限时域差分法进行数值模拟,通过建立不同煤 岩组合的几何模型,利用正演数值模拟,分析地质雷达探测煤岩界面的适用性。 最后,结合煤矿现场环境,利用 LTD-2100 雷达进行井下现场实验验证,并对实 验结果进行解释。结果表明,设置 400M 天线的 GPR 系统能够满足绝大多数煤 岩界面识别的应用, 实际应用中可以采取让地质雷达随采煤工人移动的方式实现 对煤岩界面的检测,或者采取提前勘探方式实现煤岩界面的检测。 (2)提取了不同种类和不同破碎程度的煤岩介质 ESR 吸收谱参数特征,并 对煤岩样品进行了工业分析和 X 射线衍射分析,通过对煤岩 ESR 吸收谱关键参 数朗德因子 g、线宽△H、线高 h 和自由基浓度 Ng 变化规律的分析,建立了不 同种类的煤岩自由基性状变化机理,论证了利用 X-Band 电子自旋共振波谱仪检 测煤岩自由基来进行煤岩识别的可行性。通过对吸收谱参数分析可以得出,煤岩 的 g 因子值在 2.002~2.003 之间;从煤岩类型来看,随着煤阶的增加,g 因子值 万方数据 II 呈下降趋势。特别是两种砂岩的 g 值均小于褐煤,但大于其它烟煤和无烟煤;g 值的变化与自由基种类和杂质原子有关。初步可以得出如下结论小分子自由基 g 值大分子自由基 g 值。通过对不同种类的煤岩样品的 ESR 线宽分析,可以得 出随着煤阶的增加,自由基的线宽度变窄。两种砂岩的线宽变化较小,但高于无 烟煤和贫煤,小于焦煤。也可以得出如下结论小分子自由基线宽大分子自由 基线宽。通过对不同种类煤岩样品自由基浓度 Ng 分析仪得出随着煤阶的自由 基浓度变大,随着破碎程度的增加,自由基浓度呈递增趋势。进而建立了煤岩自 由基性状的变化特性规律和机理特征。 (3)提出了一种改进的基于最小类内聚合度 IOTSU 雷达图像识别算法。传 统的 OTSU 算法只考虑目标和背景的类间方差 s,tD , 没有考虑到每一类包含的分 类信息,为了使 GPR 图像煤岩界面识别更准确,因此我们引出了最小类内聚合 度 inner s,tD , 该值越小, 表明背景和目标的类内像素内聚性越好。 在传统 OTSU 算法基础上,结合最大类间方差和最小类内聚合度提出了一种 IOTSU 算法,其 阈值选择指标为 inner ,D s,t /s,t s s tD 。当 , s st 取得最大值时所对应的 , s t,就 是二维分割的最佳阈值。实验结果表明,本文提出的方法可以有效地提高煤岩界 面识别准确率,分类识别准确率可以达到 96.3,并可有效去除为边界,目标一 致性较好。 (4)利用煤岩的自由基性状特征的差异,提出了一种 ESR-SVM-CID 的煤 岩识别方法, 解决了煤岩性状特征相近或者不含放射性元素的砂岩的煤岩识别问 题。 针对地质赋存条件复杂, 煤层含矸石情况, 提出了一种基于 ESR-SVM-CGID 的煤、矸石和岩的识别方法,解决了复杂煤层煤岩识别的问题。通过对 100 组不 同的煤岩介质的 ESR 吸收谱的主要特征参数提取,作为特征数据集,然后利用 SVM 建立了分类模型,提出了一种 ESR-SVM-CID 煤岩识别方法。实验结果表 明,煤岩介质的分类准确率可以达到 100,对不同的煤种和岩进行分类,准确 率也可以达到 100,对不同种类烟煤按高、中、低阶进行分类,准确率可以达 到 83.3。另外,对于含有矸石的复杂煤层,提出一种 ESR-SVM-CGID 方法, 也可以做到 100识别。结果表明,本文所提的煤岩识别方法,不仅可以准确识 别出不同的煤岩介质,另外还可以作为煤种分类检测的一种有效方法。煤岩介质 识别方法虽然不直接给出煤岩界面信息, 但是有效的识别煤岩介质可以作为判断 采煤机滚筒截割状态 (割煤/割岩) 的判断依据, 间接达到识别煤岩界面的目的, 从而为采煤机的滚筒自动调高提供判据。 该论文有图 50 幅,表 14 个,参考文献 190 篇。 关键词关键词煤岩识别;地质雷达;电子自旋共振;自由基 万方数据 III Abstract Underground intelligent and unmanned coal mining technology is the future development goal of China’s coal industy. The key technological breakthrough areas identified in the 12th Five-Year Plan of National Energy Science and Technology 2011-2015 include unmanned mining technology. It is pointed out that it is necessary to focus on the automatic identification technology of coal-rock characters and so on. The 13th Five-Year Plan for Safety production in Coal Mine 2016-2020 clearly pointed out promote the automation, inatization and intelligent transation of coal mine mechanization, and gradually realize the coal mine underground no-man working face. In addition, the National key basic research program 973 Plan 2014- 2018 “the key basic research of unmanned mining in deep dangerous coal seam“, which regards “the principle and realization of on-line identification of coal and rock characters and the perception of mining state“ as a key sub-subject. It is pointed out that on-line identification of coal and rock properties is the most important ination basis for realizing autonomous control of unmanned mining equipment in deep coal seam, and is the basis for realizing automatic height adjustment of shearer drum. Therefore, it is of great significance to study the identification of coal-rock interface for realizing intelligent and unmanned coal mining in fully mechanized mining face. Coal-rock identification has always been a research hotspot, although many scholars at home and abroad have put forward a large number of solutions, but these s basically stay in the feasibility of the . At the stage of discussion, there is no effective solution to study the characteristics of coal and rock. In this paper, taking coal-rock medium as the research object, taking the electromagnetic characteristics and the free radical characteristic parameter difference of coal-rock as the research contents, the related research on coal-rock character identification is carried out by using electromagnetic wave technology, and the following work is completed. 1 Taking the coal-rock interface as the research object, the coal-rock interface model with electromagnetic parameters is constructed according to the characteristics of coal-rock geological occurrence conditions, and the and applicability of coal- rock interface identification based on electromagnetic wave technology are explored. According to the three-layer scattering model of coal-rock interface, taking into 万方数据 IV account the difference between coal and rock electrical properties and the attenuation characteristics of electromagnetic wave propagation, the geological radar equation of coal-rock interface is established, which provides theoretical support for GPR detection of coal-rock interface. It also provides the basis for antenna selection. Different coal- rock combination have been carried out. The applicability of electromagnetic detection under different frequency of GPR antenna and different geological conditions is studied in this paper. Considering the complex variability of coal mine geology, the finite difference time domain FDTD is used for numerical simulation, and the geometric models of different frequencies, different dielectric constants and different coal-rock combinations are established. The applicability of geological radar to detect the coal-rock interface is analyzed by forward numerical simulation. Finally, combined with the field environment of coal mine, the LTD-2100 radar is used to verify the underground physical experiment, and the experimental results are explained. The results show that the GPR system can satisfy the application of the vast majority of coal and rock interface identification. In practical application, the detection of coal rock interface can be realized by means of ground penetrating radar moving with coal worker, or advanced exploration can be adopted. 2 The parameters of ESR absorption spectrum of different kinds of coal and rock media were extracted, and the industrial analysis and X-ray Fluorescence XRF analysis of coal and rock samples were carried out. Based on the analysis of the key parameters of ESR absorption spectrum of coal and rock, such as Langde factor g, line width △H, line height h and free radical concentration Ng, the mechanism of free radical change of different kinds of coal and rock is established. The feasibility of using X-band electron spin resonance spectrometer to detect the free radicals of coal and rock is preliminarily verified. Through the analysis of absorption spectrum parameters, it can be concluded that the value of g factor is between 2.002 and 2.003, and the value of g factor with the increase of coal rank, showing a decreasing trend. Especially, the g value of the two sandstones is lower than that of lignite, but larger than other bituminous coal and anthracite, and the change of g value is related to the kinds of free radicals and impurity atoms. A preliminary conclusion can be drawn as follows small molecular free radical g value macromolecular free radical g value. By analyzing the ESR linewidth of different kinds of coal and rock samples, it can be concluded that with the increase of coal rank, the linewidth of free radical becomes narrower. The linewidth variation of the two sandstones is small, but higher than anthracite and lean coal, 万方数据 V smaller than coking coal. It can also be concluded that the free baseline width of small molecules is larger than that of macromolecules. According to the Ng analyzer of free radical concentration of different kinds of coal and rock samples, the free radical concentration increases with the increasing of the degree of fragmentation and the increasing of the free radical concentration of the coal rank. 3 An improved maximum inter-class variance algorithm IOTSU for GPR image recognition of coal-rock interface is proposed. The traditional OTSU algorithm only considers the inter-class variance of the object and background. The classification ination contained in each class is not taken into account, and in order to make the coal-rock interface recognition more accurate in GPR images, it should also consider the minimum pixel distance within the class, so we present the minimum intra-class variance inner s,tD ,the smaller the value, the better the intra-class pixel cohesion of the background and target. By changing the threshold , s t enable inner s,tD gets the minimum value. Therefore, combined with the maximum inter-class variance and the minimum intra-class variance, an improved IOTSU algorithm is proposed in this paper. The threshold selection index is as follows inner ,D s,t /s,t s s tD . When , s s t gets the maximum , s t, the value is the optimal threshold for two-dimensional segmentation. The experimental results show that the proposed in this paper can effectively improve the recognition accuracy of coal-rock interface,the accuracy of classification and recognition can reach 96.3,and it also has a strong adaptability. 4 Based on the difference of free radical ESR character of coal and rock, a coal- rock identification based on ESR-SVM-CID is proposed, which solves the problem of coal-rock identification with similar coal-rock characteristics or no radioactive elements of sandstone. The main characteristic parameters of the ESR absorption spectrum of 100 groups of different coal and rock media were extracted, as the characteristic data set, and the original data were pre-processed. Then the classification model is established by SVM, and the classification test is carried out. The experimental results show that the classification accuracy of coal and rock media can reach 100. The accuracy of four classification of different coal types and rocks can also reach 100, and the accuracy of classification of different bituminous coals can reach 83.3 according to high, middle and low grade bituminous coals. In addition, ESR-SVM can also be used to identify complex coal seams which contain gangue,and the accuracy of classification can reach 100.The results show that the ESR-SVM-CGID coal-rock identification proposed in this paper can not only 万方数据 VI identify different coal-rock media under the condition of similar electrical parameters of coal-rock, but also have a good universality. In addition, it can also be used as an effective for coal classification and detection. The does not directly give the coal-rock interface ination, but the effective identification of the coal-rock medium can be used as a judgment basis for judging the cut-off state of the coal-mining machine coal-cutting/ rock-cutting and indirectly reach the purpose of identifying the coal-rock interface, so as to provide a criterion for automatic height-up of the drum of the coal mining machine. Keywords coal-rock identification; ground penetrating radar; electron spin resonance; free radical 万方数据 VII 目 录 目 录 摘要Ⅰ 目录Ⅶ 图清单Ⅺ 表清单XIV 变量注释表XV 1 绪论 摘要Ⅰ 目录Ⅶ 图清单Ⅺ 表清单XIV 变量注释表XV 1 绪论1 1 1.1 研究背景与意义1 1.2 国内外研究现状3 1.3 研究存在的主要问题10 1.4 研究内容与创新之处12 1.5 论文结构与小结14 2 GPR 与 ESR 电磁波相关原理和技术16 2 GPR 与 ESR 电磁波相关原理和技术16 2.1 引言16 2.2 电磁波基本理论16 2.3 界面电磁波反射和折射20 2.4 电磁波在煤岩介质中的传播特性23 2.5 电子自旋共振基本原理27 2.6 本章小结32 3 基于 GPR 的煤岩界面探测方法与适用性研究33 3 基于 GPR 的煤岩界面探测方法与适用性研究33 3.1 引言3