煤矿热动力灾害主要分为瓦斯爆炸和煤自燃 2 种情况，在采空区 2 种灾害并非单独存在。煤自燃产生的一氧化碳含量较低，需要高灵敏度在线监测，而瓦斯爆炸预警需要满足快速响应和大 量程。 可调谐激光吸收光谱技术(TDLAS)采用窄线宽激光器，选择性更好，因此适用于复杂的井 下环境。 但面对复合灾害监测，一氧化碳的特征吸收峰在近红外波段线强较弱，容易被其他井下背 景气体(乙烷、乙烯、二氧化碳、水汽)谱线覆盖，造成误报警。 因此选择 2 330 nm 波段的一氧化碳 吸收峰，该波段一氧化碳吸收峰强度高，可以实现高灵敏度检测，其次水汽在该波段的吸收峰较弱，只要识别甲烷和一氧化碳即可避免干扰。 根据高分辨率分子透射吸收数据库 HITRAN，可查询并 计算该波段一氧化碳特征峰的吸收系数，及相邻的甲烷特征峰的吸收系数。 基于可调谐激光吸收 光谱技术 TDLAS，采用 2 330 nm 激光器，相应波段小型气室和锁相模块研制一氧化碳传感器。 测 试该传感器性能，结果表明，一氧化碳的1σ检测限为2.14×10-7，3σ检测限为6.4×10-7。 测试0~ 5×10-4 内的线性度优于 2%。 充入不同谐波信号气体，控制流速为 10 mL / min，刷新时间 2 s，测试 响应时间 T90 为 6 s。 为实现井下复杂背景下的一氧化碳准确检测，笔者提出一种组分智能识别方 法。 增加激光器波长扫描范围，扫描一氧化碳和其相邻的甲烷谱线，同时获得背景气的组分和谐波 信号信息，甲烷检测限为 1×10-6 ，与 HITRAN 数据库计算结果相符。 充入混合气进行验证，测试结 果表明，该方法可以实现一氧化碳和甲烷组分识别，避免井下复杂气体作为背景的交叉干扰，同时 可以实现对一氧化碳气体的快速、高灵敏度测量，满足复合灾害监测的需求。

The thermodynamic disasters of coal mines are mainly divided into two kinds:gas explosion and coal sponta⁃ neous combustion，which are not independent in the mine goaf. In the case of coal spontaneous combustion，the car⁃ bon monoxide concentration is low，and the high sensitivity carbon monoxide sensor is needed. In the case of gas explosion warning，the methane sensor with fast response and large range is needed. TDLAS(Tunable diode laser absorption spectroscopy) with narrow line⁃width laser has better selectivity，so it is suitable for coal mine environment with complex background gas. However，in the face of the monitoring of multiple disasters，the absorption line strength of carbon mon⁃ oxide in the near⁃infrared band is relatively weak，and it is easy to be covered by the absorption line of other back⁃ ground gases in the goaf(ethane，ethylene，carbon dioxide，water vapor)，leading to false alarm. Therefore，the absorption line of carbon monoxide at 2 330 nm was selected. The line strength of carbon monoxide absorption line in this band is high，and high sensitivity detection can be achieved. Secondly，the absorption line of water vapor in this band is weak， so interference can be avoided as long as methane and carbon monoxide are identified. According to the high⁃resolu⁃ tion transmission molecular absorption database(HITRAN)，the absorption coefficients of the carbon monoxide character⁃ istic absorption peaks in this band and the absorption coefficients of the adjacent methane characteristic absorption peaks were searched and calculated. Based on TDLAS，a carbon monoxide sensor was developed by using a 2 330 nm laser， a compact chamber and a phase⁃locked module. The results show that the 1σ detection limit of carbon monoxide is 2.14×10-7 and the 3σ detection limit is 6.4×10-7. The linearity in the measurement range of 0-5×10-4 is better than 2%. Gas with different concentrations was filled into the sensor，the flow rate was controlled to be 10 mL / min，and the sensor re⁃ sponse time T90 was 6 s after testing. In order to realize the accurate detection of carbon monoxide in the background of mixed gas in goaf，a component intelligent identification method was proposed. The wavelength scanning range of the laser was increased. Carbon monoxide and its adjacent methane spectrum lines were scanned simultaneously， and the composition and concentration information of background gas were obtained at the same time. The detection limit of methane was 1×10-6，which was consistent with the calculation results of HITRAN database. Mixed gas was used for verification. The test results show that the method can realize the identification of carbon monoxide and methane compo⁃ nents，avoid the cross interference of complex gases as background，and realize the rapid and high sensitivity measure⁃ ment of carbon monoxide gas，which can meet the needs of goaf disaster monitoring.