This paper discusses the problem of selecting the working spectral points in the method of selective spectral recording, used to detect contaminants in air by means of differential optical absorption spectroscopy on acousto-optic spectrometers with random spectral addressing. Reducing the number of recorded points of the spectrum, which allows the measurement time to be reduced by an order of magnitude, can simultaneously decrease the selectivity of the analysis, and this presents a problem if the composition of the impurity substances is unknown. A partial solution of the problem is proposed that consists of a preliminary study of the variations of the spectra of the gas mixture, and this makes it possible to statistically distinguish the main variable components in the composition of the gaseous mixture and to identify them with definite gases. Versions are proposed for the construction of adaptive systems based on acousto-optic spectrometers and using the technique developed here.
© 2013 Optical Society of AmericaPDF Article