Near infrared spectrum information extraction method based on canonical correlation coefficients

Abstract

The invention discloses a near infrared spectrum information extraction method based on canonical correlation coefficients. The method comprises the steps that 1, original near infrared spectra and chemical values of tobacco leaves are acquired; 2, the abnormal spectra in the original near infrared spectra and the abnormal values in the chemical values are eliminated; 3, the near infrared spectrum is partitioned into a plurality of spectrum sections, and canonical correlation analysis is conducted on the near infrared spectrum and the chemical value in each spectrum section to calculate the canonical correlation coefficients; 4, a threshold value is selected, and PLS modeling is conducted on the wavelength points where the canonical correlation coefficients are larger than the threshold value. According to the near infrared spectrum information extraction method based on the canonical correlation coefficients, the explanation capacity of the near infrared spectrum can be improved, the multicollinearity of near infrared spectrum information can be eliminated, and the stability of predicting a chemical component model through the near infrared spectrum can be improved.

Description

technical field [0001] The invention relates to the technical field of tobacco leaf component analysis, in particular to a method for extracting near-infrared spectral information based on typical correlation coefficients. Background technique [0002] Near-infrared spectroscopy is widely used in the field of quantitative rapid detection, and provides a good method for obtaining basic material data for on-site quality control. The near-infrared spectrum belongs to the electromagnetic wave in the 850-2500nm band, and different types of spectra reflect molecular information at different sample levels. Near-infrared spectral information approximately conforms to the Lambert-Beer formula, which provides a theoretical basis for extracting composition and structure information of detected substances from near-infrared spectral information. [0003] However, in the actual near-infrared testing process, it is faced with complex changes in the use environment, annual differences in ...

AI Technical Summary

Problems solved by technology

[0005] First, the material information contained in the near-infrared spectral signal is very weak, and the effective spectrum difference of the sample in some spectral regions is very small, which cannot clearly provide the measured component information of the sample;

[0006] Second, in some bands, there is no linear correlation between the near-infrared spectral information of the sample and the measured composition or properties of the sample. Once the linear partial least squares method is selected for the modeling method, the accuracy of the model will be reduced;

[0008] Fourth, due to the noise of near-infrared spectroscopy instruments, in some bands, the sample spectral signal-to-noise ratio (SNP) is low, and the spectral quality is poor. If these noises are not filtered out, the model is not robust;

[0009] Fifth, if full-spectrum modeling is used, once the external environmental factors change, it will interfere with the absorbance of the near-infrared spectrum, making the detected sample an outlier;

[0010] Sixth, when the number of wavelength variables is too large, the model calculation is complicated and the model accuracy is reduced

This method is very practical in near-infrared wavelength selection, but it is only suitable for simple linear systems

[0014] (2) Uninformative variable elimination method (UVE): This method not only uses the absolute value of the regression coefficient, but also takes into account the variance of the regression coefficient, and integrates noise, spectrum and concentration information. It is intuitive and practical, but this method requires A certain sample size, when the sample size has not accumulated to a certain extent, the effect of improving accuracy is not obvious

[0016] (4) CARS-Monte-Carlo-sampling algorithm: When this method is actually used, the stability of the sample is very high. If there is a problem with the benchmark data of the model or there is a problem with the representativeness of the actual sampling spectrum and sampling, There will be obvious errors in the construction of the model

Images