煤自燃多组分指标气体TDLAS痕量检测技术基础研究

The determination of the risk degree of coal spontaneous combustion is a worldwide problem. Gases are the key indicators of coal fire hazard detection. In the oxidation and pyrolysis process of coal spontaneous combustion, it simultaneously produces constant, minor and trace index gases . Existing infrared detection techniques has poor selectivity, limited types of gas detection. Chemical sensors possess cross-sensitive, zero drift problems. GC is complex and time consuming. These methods are difficult to achieve accurate, real-time online detection. Tunable diode laser absorption spectroscopy technology has the advantages of high selectivity, high sensitivity, high precision and real-time online detection. It can simultaneously detect a variety of gases. It is the trend of gas detection. The topic for the key scientific issues of the TDLAS trace detection of multi-component index gases, which intends to adopt the sparse decomposition method and the triple Lorentz function model, the problem can be solved for the separation of self overlapping spectrum lines of olefin gas. By establishing the correction model of MPSO-ELM, the crossing interference of hydrocarbon gas absorption lines is excluded. The background noise of waiting for the tested gas absorption spectrum lines is deducted by arithmetic mean - wavelet filtering method. The high sensitivity trace detection of multi-component index gases are realized by the technique of phase sensitive demodulation and low-pass filtering. The technology provides an advanced tool and technical support for the determination of the risk degree of coal spontaneous combustion, the realization of the early prediction of coal spontaneous combustion and the scientific decision-making, and to ensure a safe rescue.

煤自燃危险程度的判定是世界性难题，气体是煤火灾害检测的关键指标。煤自燃时，氧化和热解同时产生常量、微量和痕量的指标气体，现有红外检测技术选择性差、检测种类受限，化学传感器交叉敏感、零点漂移，气相色谱使用复杂、时效性差，这些方法均难以实现精确、实时在线检测。可调谐半导体激光光谱分析技术具有高选择性、高灵敏度、高精度及实时在线测量等优点，可同时检测多种气体，是气体检测技术的发展方向。针对多组分指标气体TDLAS痕量检测存在的关键科学问题，采用稀疏分解和三重洛伦兹函数法，解决烯烃类气体自身重叠谱线分离的难题；建立MPSO-ELM修正模型，排除烃类气体吸收谱线的交叉干扰；采用算术平均-小波滤波法，扣除待测气体吸收谱线的背景噪声；利用相敏检波和低通滤波技术，实现多组分指标气体的高灵敏度痕量检测。该技术为判定煤自燃危险程度，实现煤自燃火灾的早期预测与科学救灾决策及确保救灾安全提供先进手段和技术支撑。

煤自燃危险程度的判定是世界性难题，气体是煤火灾害检测的关键指标。煤自燃时，氧化和热解同时产生常量、微量和痕量的指标气体，现有红外检测技术选择性差、检测种类受限，化学传感器交叉敏感、零点漂移，气相色谱使用复杂、时效性差，这些方法均难以实现精确、实时在线检测。可调谐半导体激光光谱分析技术具有高选择性、高灵敏度、高精度及实时在线测量等优点，可同时检测多种气体，是气体检测技术的发展方向。针对多组分指标气体TDLAS痕量检测存在的关键科学问题，基于HITRAN2016数据库获取的数据，筛选出各组分气体吸收谱线的最佳中心波长，并给出了煤自燃多组分指标气体吸收系数的计算方法；将建立的光路系统模型导入Light Tools仿真分析，优化了TDLAS光路系统，提高了TDLAS技术的分辨率和灵敏度；提出了温度与压力对气体浓度影响的归一化数据处理分析方法，建立了基于SA-PSO-SVM的TDLAS检测系统误差温压补偿方法，与其他方法相比，SA-PSO-SVM补偿方法预测精度以及鲁棒性最佳；针对多组分混合气体中气体谱线之间的交叉干扰，提出了多重Lorentz函数分解法和多谱线拟合算法两种检测方法，多重Lorentz函数分解法是采用一个四重Lorentz函数模型去描述C2H4气体的吸收系数，测量精度高，数据处理速度快，在多组分混合气体测试中，该方法可有效地实现C2H4气体吸收谱线的分离及其浓度的精确测量，对C2H4气体浓度的测量误差小于5ppm。多谱线拟合算法是结合波长调制技术，将测量的WMS-2f/1f信号与仿真的WMS-2f/1f信号进行多谱线拟合，根据提取出的谱线信息实现对温度、压强和乙烯浓度的同时检测。通过气体检测实验，验证了检测系统的有效性，系统在检测灵敏度、稳定性、重复性等性能指标上表现良好，可以满足高精度气体检测的需求。该技术为判定煤自燃危险程度，实现煤自燃火灾的早期预测与科学救灾决策及确保救灾安全提供了先进手段和技术支撑。