Coal spontaneous combustion (CSC) has been a global hazard for decades, causing significant losses. Hydrocarbon gases, including carbon monoxide (CO), carbon dioxide (CO2), methane (CH4), ethylene (C2H4), acetylene (C2H2), and oxygen (O2), have proved to be good inhibitors for forecasting CSC. However, the cross-interference and absorption spectrum overlaps prevent their practical applications. This study simulates the refined distribution of the absorption lines of these index gases in the infrared spectral range to solve these problems. By selecting the optimal absorption lines for each gas, their detection performance was experimentally tested, and the results were analyzed using the Allan variance method. The results reveal that the optimal absorption lines are centered at 1566.64, 1572.32, 1653.72, 1626.34, 1530.37, and 760.65 nm for CO, CO2, CH4, C2H4, C2H2, and O2, respectively. Relative detection errors are 0.62, 0.51, 3.06, 4.20, 0.58, and 1.96%, and the detection limits are 3.47×10−6, 4.56×10−6, 0.53×10−6, 2.85×10−6, 0.33×10−6, and 1581×10−6, respectively. The detection sensitivity and comprehensive detection accuracy were significantly improved. This study will provide a basis for solving the problem of the cross-aliasing interference between index gases for bituminous CSC.