Qualitative analytical method of infrared spectrum

Abstract

The invention discloses a qualitative analytical method of an infrared spectrum. The method comprises the following steps of: primarily screening components on a to-be-detected sample spectrogram from a spectral database by employing an ENet method; then establishing a CLS model for loop iterative calculation to remove components, the CAE of which is less than 0, to secondarily screen the components; and finally, fitting YAC according to second screening spectral data, and calculating and judging the sizes of PA1 and PA2 to screen the components for the third time. In the whole processes, a variable selection ability of the ENet method is combined, and the infrared spectrum is quickly and accurately recognized and analyzed qualitatively through a loop iterative CLS process and a UPAs process. Meanwhile, the method needs not to establish a model in advance and also can solve the problem of severe multicollinearity of the spectral components in a multi-component infrared spectrum.

Description

technical field [0001] The invention relates to the field of infrared spectrum analysis, in particular to a qualitative analysis method of infrared spectrum. Background technique [0002] Infrared spectral analysis is an important method for substance qualitative. Its analysis can provide a lot of information about functional groups, and can help determine some or all molecular types and structures. Its qualitative analysis has high characteristics, short analysis time, and requires a small amount of samples. , No damage to the sample, convenient measurement and other advantages. Qualitative analysis of infrared spectroscopy is mainly used for qualitative identification of substances, that is, to determine the identity of unknown substances by comparing the spectra of unknown samples with known reference sample sets. With the development of qualitative analysis methods of infrared spectroscopy, especially Fourier transform infrared spectroscopy, there are already many metho...

AI Technical Summary

Problems solved by technology

With the development of qualitative analysis methods of infrared spectroscopy, especially Fourier transform infrared spectroscopy, there are already many methods in the prior art for qualitative analysis of Fourier transform infrared spectroscopy: such as the classical least squares regression method (CLS) , Partial Least Squares (PLS), Artificial Neural Network Algorithm (ANN), etc., there are certain defects in the qualitative analysis of infrared spectra based on these methods

For example, when CLS is used for component identification, it is first necessary to select the absorbance data of all spectra in the investigated band, but all spectral data can eventually affect the identification results of the algorithm. Multicollinearity problem, and when using PLS modeling for qualitative analysis, the appearance of new components will reduce the accuracy of identification; in recent years, the artificial neural network algorithm has developed rapidly, but the algorithm requires a complicated training process, and a single artificial The neural network algorithm can only qualitatively identify a single component, and cannot achieve rapid qualitative identification analysis

[0003] At the same time, with the increase of the types of infrared spectra in the infrared spectrum database, the collinearity of the absorption spectrum characteristics of the components increases, and the existing qualitative analysis methods of infrared spectra are difficult to solve such problems

Therefore, for real-time online applications, it is still difficult to achieve accurate and fast qualitative analysis of the infrared spectra of the samples to be tested from the large database of infrared spectra

Images