赋存CO气体的煤层自燃D-S证据融合预测研究

发布时间：2018-01-30 10:13

本文关键词： 煤自燃 CO赋存 BPA D-S理论预测　出处：《北京科技大学》2017年博士论文　论文类型：学位论文

【摘要】：我国煤炭自燃灾害非常严重,煤层自燃不仅直接造成资源的浪费,给矿山企业带来巨大的经济损失,而且还严重威胁企业人员的生命安全、破坏矿井设备设施、危害周边自然环境,科学、及时、准确地预测煤层自燃火源对企业安全生产尤为重要。煤自燃预测的一个关键指标是CO气体,但如果煤层本身赋存有CO气体,在开采过程中解吸逸散的CO气体会干扰矿山煤自燃预测预报工作。本文基于上述问题,选取开滦集团下属林南仓矿为研究对象,针对该矿煤层是否赋存有CO气体的争论,通过现场测试和实验室实验证实了煤层赋存有CO气体,分析了CO气体赋存成因的影响因素,并在此基础上提出一种以CO为核心指标,辅以煤温及C_3H_8、C_2H_4指标气体,提取监测数据的信息特征,建立以基本信任分配函数(BPA)为基础的煤自燃预测指标体系,运用多源信息融合D-S(Dempster/Shafer)证据理论实现了赋存CO气体的煤层自燃预测。1、采用现场钻孔取气实验、煤样真空解吸实验并运用协整理论求证了煤层赋存有原生CO气体,并从矿井地质结构、煤岩大分子结构、煤岩吸附性能、煤岩渗透性能及煤岩孔隙结构特征等方面分析了其对林南仓矿CO气体赋存成因的影响。2、采用智能煤升温氧化箱，，有效模拟矿井自然发火过程,总结分析了气体指标参数与煤温的对应关系,选取CO、C_2H_4、C_3H_8为煤自燃预测的标志气体,并基于CO气体浓度变化特征将煤自燃前期划分为赋存气体解吸期、氧化低危险期、氧化中危险期、氧化高危险期、氧化极高危险期5个阶段,利用模糊隶属度(Fuzzy Membership Function)与IS区间集两种方法构建了以基本信任分配函数(BPA)为基础的煤自燃预测指标体系。3、运用多源信息融合D-S证据理论,分别就采煤工作面和采空区进行了煤自燃D-S证据理论融合预测模拟验证,并针对多源信息融合中证据冲突问题,提出了一种基于“相对差异”的证据冲突度量方法。模拟结果显示,本文给出的煤自燃预测方法以特征识别保障了价值信息的有效保留,多源融合增加了预测结果的准确性,同时为适应井下复杂情况提供了有效的决策依据。
[Abstract]:Coal spontaneous combustion disaster is very serious in our country. Coal seam spontaneous combustion not only directly causes waste of resources and brings huge economic losses to mining enterprises, but also seriously threatens the safety of enterprise personnel and destroys mine equipment and facilities. It is very important for enterprises to predict the spontaneous combustion source of coal seam in a scientific timely and accurate manner which endangers the surrounding natural environment. A key index of coal spontaneous combustion prediction is CO gas. However, if there is CO gas in the coal seam itself, the CO gas desorbed and dissipated during the mining process will interfere with the prediction and prediction of spontaneous combustion of mine coal. This paper selects Linnancang Mine of Kailuan Group as the research object, aiming at the argument of whether there is CO gas in the coal seam of this mine, and confirms the existence of CO gas in the coal seam by field test and laboratory experiment. Based on the analysis of the factors influencing the occurrence of CO gas, a kind of index gas, which is based on CO as the core index and supplemented by coal temperature and C _ 3H _ 8C _ 2H _ 2H _ 4 index gas, is put forward. The information features of the monitoring data are extracted and the prediction index system of coal spontaneous combustion is established based on the basic trust distribution function (BPA). Based on the evidence theory of multi-source information fusion D-Sine Dempster / Shafer, the prediction of spontaneous combustion of coal seam with CO gas is realized. The coal sample vacuum desorption experiment and cofinishing theory are used to verify the existence of primary CO gas in coal seam, and from the geological structure of coal mine, coal and rock macromolecular structure, coal and rock adsorption performance. The influence of coal and rock permeability and pore structure on the genesis of CO gas in Linancang Mine is analyzed. The intelligent coal heating and oxidation box is used to simulate the spontaneous combustion process effectively. The relationship between the gas index parameters and the coal temperature is summarized and analyzed, and the COG _ Cs _ 2H _ 4 / C _ 3H _ 8 is selected as the mark gas for the prediction of the spontaneous combustion of coal. Based on the change of CO concentration, the coal spontaneous combustion period is divided into five stages: the desorption period of existing gas, the period of low danger of oxidation, the period of middle danger of oxidation, the period of high danger of oxidation, and the period of extreme danger of oxidation. Using fuzzy Membership function) and is interval set, the basic trust assignment function (BPA) is constructed. Based on the prediction index system of coal spontaneous combustion. 3. Based on the D-S evidence theory of multi-source information fusion, this paper simulates the prediction of D-S evidence theory fusion of coal spontaneous combustion in coal mining face and goaf, and aims at the conflict of evidence in multi-source information fusion. A measure method of evidence conflict based on "relative difference" is proposed. The simulation results show that the prediction method of coal spontaneous combustion based on feature recognition ensures the effective retention of value information. Multi-source fusion not only increases the accuracy of prediction results, but also provides an effective decision basis for the complex underground situation.
【学位授予单位】：北京科技大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TD752.2

【参考文献】