基于光声光谱煤矿瓦斯检测系统的研究及应用

【作者】 赵南；

【导师】 郑德忠；

【作者基本信息】 燕山大学 ， 仪器科学与技术， 2014， 博士

【摘要】 煤矿瓦斯事故是我国煤矿安全生产的头号大敌。我国能源结构呈现富煤少油的特点，且煤矿地质构造复杂，高瓦斯矿井所占比重大，造成了我国煤炭瓦斯事故呈现长期、不可避免、易发而严峻的形势。煤矿瓦斯防治工作是我国安全生产的重中之重。煤矿瓦斯监测及瓦斯事故预测预警是保障我国煤矿安全生产的重要有效手段，也一直是能源领域专家学者研究的重要课题。本文通过对光声光谱气体检测原理的系统研究，深入分析并建立了基于光声光谱的煤矿瓦斯检测系统，提出一种瓦斯爆炸预报警的新思路，并对其进行了实验验证。主要研究内容如下：通过对光声光谱气体检测原理深入系统的研究，分析光声光谱检测煤矿瓦斯浓度的可行性。针对煤矿瓦斯成分的特点，利用HITRAN和HITEMP数据库，通过对甲烷以及矿井环境气体中水蒸气、二氧化碳、一氧化碳、一氧化氮、硫化氢、二氧化硫、二氧化氮等吸收谱线的比较分析，选择1653.79nm吸收谱线进行甲烷的光声信号探测。通过对光声信号产生机理和光声腔内声场分布深入的研究，论证了光声信号与光声腔结构及其工作模式的内在联系，指出光声信号与一阶纵向圆柱形谐振式光声腔结构尺寸间的关系，为设计新型光声腔提供了理论依据。在充分考虑了实际工作过程中光声腔内部存在损耗的基础上，结合降噪措施设计了新型长度可调一阶纵向反馈谐振式光声腔。结合自主设计的光声腔，设计实现了基于光声光谱的煤矿瓦斯检测系统，并对光声腔的特性参数进行测量，得谐振频率为1003Hz、品质因数为35.05、腔常数为4194.59Pa cm W-1，同时测得系统检测灵敏度为2.78×10-6，信噪比为39.5。通过对多种缓冲气体及待测气体的实验，分析了缓冲气体种类、测量温度、激光器输出功率等因素对光声腔特性参数、光声信号以及系统检测灵敏度的影响。在对瓦斯爆炸条件分析的基础上，提出了瓦斯爆炸预报警的新思路，即同时满足瓦斯爆炸三个条件中任意两个时报警。建立了基于神经网络模式识别算法的煤矿瓦斯爆炸预报警网络模型，利用冀中能源峰峰煤矿的监测数据对瓦斯爆炸预报警网络进行训练，运用混淆矩阵、ROC曲线、MSE曲线、误差直方图等分析手段对网络性能逐一分析，得预报警准确率为93.8%。通过对煤矿井下实际报警前后环境数据变化情况的模拟，运用光声光谱瓦斯监测系统对瓦斯浓度进行实时监测，瓦斯爆炸预报警网络进行预报警判断，所得报警时间比矿井实际报警时间提前更多还原

【Abstract】 Coal mine safety accidents is the first enemy of coal mine safety production atpresent. The characteristics of energy structure in our country is rich in coal but poor in oil.The fact that geological structure of coal mine is complex and gas mine accounts for morethan major is the main reason of coal gas accident in our country being in a long-term,inevitabe, and serious situation. Coal mine gas prevention and control work is the toppriority of safety in China. Methane concentration detection and fault forecasting are theimportant and effective safeguards of the production safety of coal mines in our country,and has always been an important subject of the experts and scholars in the field of energyresearch. In this article, we studied the photoacoustic spectroscopy technology principle ofgas detection system, deeply analysied and finally established a gas detection systembased on photoacoustic spectroscopy technology, and proposed a new method in gasexplosion pre-alarm. Through the experiments, the gas detection and gas explosionpre-alarm system were demonstrated respectively. The main research content is as follows.Through depth study of the photoacoustic spectroscopy gas detection principle,analyzes the feasibility of coal mine gas concentration of photoacoustic spectra. Accordingto the characteristics of the mine gas composition, based on HITRAN and HITEMPdatabase, analysis the methane absorption spectrum characteristic. Gases which are inmine air and methane gas such as water vapor, carbon dioxide, carbon monoxide,hydrogen sulfide, sulfur dioxide, nitrogen dioxide, nitric oxide and some other absorptionlines are analyzed and compared, finally selected1653.79nm methane absorption lines forphotoacoustic signal detection.Based on the depth study of the photoacoustic signal generation mechanism and theacoustic field distribution inside the photoacoustic cell, demonstrated the internal relationsabout photoacoustic signal, photoacoustic cell structure and its working mode. Futherpresents the relationship between photoacoustic signals and first-order vertical cylindricalresonant photoacoustic cell structure and size. All above provides a theoretical basis forthe design of novel photoacoustic cell. Based on fully consideration the internal loss of thephotoacoustic cell in actual working process, combined with the measure of noisereduction, designed the new length-tunable first-order longitudinal feedback resonantphotoacoustic cell. Combined with self-designed photoacoustic cell, designed and implemented the minegas concentration measurement system which based on photoacoustic spectroscopy, andthe characteristics of photoacoustic cell parameters were measured. Following results wereobtained: resonant frequency was1003Hz, quality factor was35.05, cell constant was4194.59Pa cm W-1, and the detection sensitivity was2.78×10-6, SNR was39.5. Formany kinds of buffer gas and measuring gas experiments, analyzes the factors such astypes of buffer gas, measuring temperature, laser output power effect the photoacousticcell performance parameters, the value of photoacoustic signal, and then the detectionsensitivity of the system.Based on the analysis of conditions in the gas explosion, presented a new idea whichis satisfied two of the three basic gas explosion conditions, and based on neural networkpattern recognition algorithms established mine gas explosion pre-alarm network model.the datas by Jizhong Energy Fengfeng mine monitored was used in network training.Using the confusion matrix, ROC curve, MSE curves, the error histogram and some otheranalysis methods analyze the network performance, and achieved the accuracy rate was93.8%. Through simulated the actual changes of the coal mine environmental data beforeand after the alarm, monitored gas concentration real-time using photoacousticspectroscopy gas monitoring system, judged pre-alarm condition by gas explosionpre-alarm network. Point out that alarm time was shorter than the actual alarm time inmine, pre-alarm effect further improved.更多还原