煤矿瓦斯监测多传感器信息融合与知识发现研究

发布时间：2018-04-02 01:33

本文选题：瓦斯监测　切入点：多传感器信息融合　出处：《中国矿业大学》2013年博士论文

【摘要】：瓦斯灾害防治仍然是我国煤矿安全工作的重中之重。综合利用布设在井下空间的各类非接触式传感设备动态采集的相关数据，对具有突出危险的工作面实现实时跟踪监测和早期诊断预警，为煤矿及时采取针对性措施，提高监控系统可靠性，防范和抑制瓦斯突出、瓦斯积聚和瓦斯爆炸等事故提供决策依据，是目前煤矿瓦斯安全监测系统亟待增强的功能目标。论文根据煤与瓦斯突出、瓦斯爆炸等危险性预测技术和预警理论，采用多传感器信息融合方法，充分挖掘瓦斯、风速、电磁辐射、声发射等各类传感数据所蕴涵的规律性知识，发挥各类传感器的优势，按照“手段多样、优势互补、相互验证、短中长期搭配”的思路，着力构建基于多传感器信息融合的瓦斯安全监测预警系统，实现对井下工作面瓦斯危险的“实时感知、准确辨识、快速响应、有效控制”。论文取得的主要研究成果如下：全面总结分析了国内外煤矿瓦斯安全动态监测手段和突出危险性评价指标的研究成果，包括瓦斯浓度、电磁辐射、声发射等传感监测技术及突出预测方法，为发挥各自优势，实现煤与瓦斯突出多传感器融合预警奠定了坚实的理论基础。通过现场调研，分析了煤矿瓦斯灾害防治实际需求，本着提高监测系统效能，降低系统资源消耗的理念，提出了瓦斯监测多传感器信息融合的目标体系、闭环工作流程、传感器选用与组织以及各种瓦斯安全动态监测传感信息融合分析理论的合理运用，从而最终确定了瓦斯监测多传感器信息融合体系总体结构，，重点研究了基于模糊专家系统的瓦斯突出预测多传感器信息决策融合方法。提出了基于时间序列相似性度量的瓦斯超限报警信号辨识方法。基于DTW距离对煤矿采掘工作面瓦斯超限报警时间序列进行了聚类分析。对所获得的7种典型的时间序列模式，基于分段形态度量方法，提取了15个特征指标，从中筛选出5个分类效能较强指标，建立了瓦斯超限报警时间序列形态特征表。在此基础上提出了一种瓦斯报警信号快速辨识算法。提出了基于时空相关分析的煤矿采掘工作面瓦斯监测数据异常自动识别技术。定性分析了工作面顺风流方向瓦斯运移存在的时空异步相关特性；确定了相关系数计算过程中涉及的异步相关最优滞后步长的计算方法和瓦斯气体涌出后在回风巷道中体积分数随时空变化的预测和反演公式；统计计算了8种原因导致的瓦斯数据异常存在的相关系数值变化区间；提出了基于时空相关分析的工作面瓦斯监测数据异常识别算法；为提高相关分析效率，提出了能表达空间拓扑信息的井下瓦斯传感器层次编码方法。提出将工作面瓦斯安全监测问题归类为专家诊断范畴。研究了瓦斯监测信息知识发现方法，提出了瓦斯时间序列聚类分析与知识提取方法；针对瓦斯监测多传感器信息决策融合专家知识库系统的设计需求，提出了瓦斯监测知识学习算法和瓦斯监测专家知识的组织存储策略。最后举例说明了基于专家系统的工作面瓦斯超限原因识别推理应用过程。
[Abstract]:Gas disaster prevention is still the priority among priorities of coal mine safety work in China. Comprehensive utilization of data layout of non-contact dynamic acquisition sensor equipment of all kinds in the underground space, with outburst dangerous to achieve real-time tracking monitoring and early diagnosis and early warning, for the coal mine to take corresponding measures to improve system reliability. The prevention and suppression of gas outburst, provide decision-making basis for gas accumulation and gas explosion accident, is currently the target function of coal mine gas safety monitoring system needs to be enhanced.
According to the coal and gas outburst, gas explosion hazard prediction and early warning theory, using information fusion method, fully tap the gas, wind speed, electromagnetic radiation, knowledge of the law of acoustic emission and other types of sensing data contains all kinds of sensors, play advantage, "according to a variety of means, complementary advantages, mutual authentication. In the short term collocation" ideas, focus on building a gas safety monitoring and early warning system based on multi sensor information fusion, realize the gas in the working face of the dangerous "real-time perception, accurate identification, rapid response, effective control. The main achievements of this study are as follows:
It analyzes the research results of domestic and foreign dynamic monitoring system of coal mine gas safety and outburst risk evaluation index, including gas concentration, electromagnetic radiation and acoustic emission sensing technology and prediction methods, to play their respective advantages, realize the coal and gas outburst early warning sensor fusion laid a solid theoretical foundation.
Through field investigation, analysis of the mine gas disaster prevention based on the actual demand, improve the efficiency of the monitoring system, reduce the system resource consumption concept, put forward the target gas monitoring system of multi-sensor information fusion, closed-loop workflow analysis, rational use of theory of selection and organization as well as various sensor gas safety monitoring sensor information fusion, and finally the multi sensor information fusion system structure of gas monitoring, focusing on the prediction of multi sensor information decision fusion method of gas outburst based on fuzzy expert system.
The similar gas measurement based on time sequence alarm signal identification method. The DTW distance of gas in coal mining work face alarm time series clustering analysis based on 7 typical time series model is obtained by segmented shape measurement method based on extraction, 15 indicators, selected 5 classification the strong performance index in the establishment of a gas alarm time series characteristics table. This paper presents a gas alarm signal identification algorithm.
The abnormal automatic identification technology of gas in coal mining work face spatio-temporal correlation analysis based on the monitoring data. The qualitative analysis of the working surface along the spatial correlation characteristics of gas migration in asynchronous wind direction; the correlation coefficient calculation method and the gas in the process of asynchronous step involves optimal lag long after pouring in ventilation roadway in volume fractional change with the time and space prediction and inversion formula; statistical analysis of abnormal gas data of 8 kinds of causes of the correlation coefficient change interval; proposed anomaly recognition algorithm of space time correlation analysis based on the monitoring data of the gas in the working face; in order to improve the efficiency of correlation analysis, put forward the underground gas sensor level encoding method can express spatial information.
The working face of the gas safety monitoring problems classified as expert diagnostic category. On the gas monitoring information of the knowledge discovery method, proposed gas time series clustering analysis and knowledge extraction method; according to the design requirements of Fusion Expert System for gas monitoring of multi sensor information decision, proposed gas monitoring knowledge learning algorithm and storage strategy for gas monitoring expert knowledge. An example of gas expert system working face recognition application process based on the causes of overloading.

【学位授予单位】：中国矿业大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：TD712

【参考文献】