Fufurin

Detection of substances with the use of their infrared spectra is carried out by means of statistical analysis of experimental and reference spectra. This analysis is conducted by calculating the correlation parameter. In this work dependences of correlation parameters of reference and noisy spectra on the signal-to-noise ratio were obtained. Dependence of substance recognition thresholds on the signal-to-noise ratio within the experimental spectra was obtained for the given probability of substance recognition. The authors consider the possibility of applying these dependences to calculation of recognition thresholds in view of the signal-to-noise ratio during the experiment.

 The authors consider numerical solution algorythms for identification and concentrations calculation of substances in the atmosphere. Instead of solving the multidimensional system of linear equations it is proposed to solve one or several one-dimensional systems. Conversion to one-dimensional problem is performed by correlation analysis. The influence of noise in the initial spectrum on errors of inverse problem solution was studied.