The remote detection of gas infrared radiation signal in high temperature kiln is carried out based on Fourier transform infrared spectrometry technique. According to industrial site conditions, a passive radiation model is built based on the principle of atmospheric radiation, and the high temperature gas transmittance in the cement-kiln is calculated. Aiming at the influence of turbulence noise on the signal to noise ratio, a data processing method for spectrum inversion of infrared interference signal is studied, and the average signal of multiple scanning interference signal is obtained when we align the interference signal on the basis of zero optical path difference. The noise and the amount of computation are reduced, and the spectral data rate is improved. Based on the HITRAN database and the high temperature reference spectrum model, the nonlinear least square fitting method is used for the spectrum line strength and corrected and transmission spectra obtained by line width correction, and the concentration of high temperature gas in cement-kiln in different absorption bands is inverted. The results show that the proposed method for on-line monitoring of high temperature gas in the cement-kiln and other industrial combustion processes is feasible and reliable.