【作者】 张杰；

【导师】 马宏伟；

【作者基本信息】 西安科技大学 ， 安全技术及工程， 2014， 博士

【摘要】 矿井瓦斯一直是我国煤矿主要的灾害形式之一，并严重困扰着煤矿的安全生产。研究矿井日常检/监测数据的有效处理及其预测预警应用，有利于拓展安全监测监控系统的功能，是提高瓦斯灾害预警能力的重要手段。本论文在分析矿井瓦斯检/监测数据特征及其集成管控模式的基础上，深入研究了基于云计算数据集成模式下的矿井瓦斯预警分析理论和方法。研究了矿井瓦斯检/监测数据的特点及其集成管控模式。分析了瓦斯检/监测数据的特征，并对于环境、人为、管理等因素影响下存在的异常数据、数据缺失问题，针对其特征进行平滑处理，使其形成完整的监测数据序列，符合监测数据整体统计特性，并构建了云计算模式下检/监测数据集成管控模式。研究了矿井瓦斯浓度变化趋势预测预警方法。在瓦斯监测数据预处理的基础上，基于时间序列分析的自回归滑动平均（ARMA）模型，建立了适用于实时监测数据的瓦斯浓度动态趋势预测预警分析模型，结合实时预测结果与所在时段瓦斯监测数据的统计特征实现了瓦斯浓度变化趋势的动态预警。研究了矿井瓦斯突出危险性预测预警方法。通过分析瓦斯实时监测数据的特征，提取反映瓦斯浓度实时变化趋势的参数、瓦斯浓度变化速率的参数以及用于表达瓦斯涌出特征的参数，结合防突检测参数，基于v-支持向量机（v-SVM）模型，构建了瓦斯突出危险性预测预警模型，结合瓦斯突出危险性预测结果与防突检测数据的统计特征，实现了瓦斯突出危险性预警。研究了矿井瓦斯预警的云计算模型架构。基于云计算的原理，构建了应用于矿井瓦斯预警分析的云计算模式的物理架构及其云计算平台模式，并将矿井瓦斯检/监测数据处理及预测预警算法予以封装，为瓦斯预警计算的程序化服务构建了云计算模式，实现了高效的预警分析。研究了云计算数据集成模式下瓦斯预警分析应用。基于所建立的瓦斯预警数学模型，将瓦斯检/监测数据处理的云计算模式应用于现场预警分析，经过实际检/监测数据的对照检验，表现出了良好的适用性和有效性。本论文研究的云计算数据集成模式下瓦斯预警分析理论和方法，适用于煤矿现场的瓦斯预警分析应用，为煤矿瓦斯灾害防治提供了新的数字化平台构建方法和手段。更多还原

【Abstract】 As one of the major hazards, mine gas seriously impedes safe production in coal mines.The study on effective analysis of mine monitoring data for the pre-warning of abnormal gassituation is important means to expand functions of the safety monitoring and control systemas well as to improve safety pre-warning. Based on the monitoring data and data integration,mine gas pre-warning under cloud computing environment is studied.Characteristics of gas monitoring data and its integration models are studied. The timesequence of gas monitoring data is re-built in a smooth process on the basis of statistics so asto treat data defects caused by sensor errors and irregularity of measurement due to variousinterferences from environment, human and management factors, thus the data is ready forcloud computing.Mathematical models of gas concentration prediction and pre-warning of abnormal gasemission is studied. By performing data pre-processing and the ARMA time sequenceanalysis, a dynamic prediction and pre-warning model for the monitoring system is putforward, and the on-line dynamic pre-warning is realized based on real-time prediction of gasconcentration and statistics of gas measurements.Hazard prediction and pre-warning of mine gas outburst is studied. By identifying thecharacteristics of monitoring data, the progressive trend and the drift rate of gas concentration,features of gas emission, and the risks of gas outburst can be judged, and gas outburst hazardprediction and pre-warning models based on v-SVM platform is constructed, and on-site riskprediction of gas outburst is realized.A model of mine gas pre-warning supported by cloud computing is studied, and thephysical structure and platform model of cloud computing suitable for gas pre-warninganalysis is put forward. Then the algorithms of data processing and gas pre-warning areencapsulated, and services of cloud computing as well as effective pre-warning analysis arerealized. The gas pre-warning under cloud computing environment is studied. Based on gaspre-warming models, the pre-warning analysis supported by cloud computing is adopted andpromoted for on-site application. Verified by systematically obtained monitoring data, itshows that the models exhibit good applicability and validity.Therefore, this study of gas pre-warning analysis theory and methods under cloudcomputing environment is applicable to mine site. It is capable to offer scientific thinking forthe digital platforms of mine gas disaster prevention and control.更多还原