Near infrared spectrum qualitative analysis method with universality

Abstract

The invention discloses a near infrared spectrum qualitative analysis method with universality. The method comprises the following steps: acquiring an infrared spectrum of each of a training set sample and a prediction set sample, constructing a cosine dictionary, solving expression coefficients of the training set sample and the prediction set sample in the cosine dictionary, carrying out the spectrum reconstruction in combination with sparseness of the expression coefficients, searching qualitatively classified effective projection direction according to reconstructed training set near infrared spectrum and training set sample classifying number, and finishing the projection transformation of the reconstructed training set infrared spectrum and the reconstructed prediction set near infrared spectrum, and then computing the Laplacian matrix of the transformed training set infrared spectrum and the transformed prediction near infrared spectrum, and transforming the classification problem into the optimization problem so as to finally finish the solving of the category function and execution of the qualitative analysis on the prediction set near infrared spectrum. The invention provides a near infrared qualitative analysis method with universality, wherein the method is high in precision and good in accuracy and is effectively suitable for small sample complex system.

Description

technical field [0001] The invention relates to the field of near-infrared instrument analysis and detection in chemical industry, food, agriculture, environment, biology and other industries, in particular to a near-infrared spectrum qualitative analysis method. Background technique [0002] Near-infrared spectroscopy is a method of qualitative analysis using the absorption intensity of sensitive groups in substances to near-infrared light of different wavelengths. In recent years, it has been widely used in agriculture, food, medicine, medicine, petrochemical and other fields. After the application of near-infrared analysis technology in various fields, the advantages are obvious, mainly in the following aspects: the analysis and test speed has been greatly improved, and the test time is usually within a few minutes; the test efficiency has been greatly improved. Simultaneously measure multiple properties of the sample; the test cost is further reduced, and the near-infrar...

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the deficiencies of the existing near-infrared spectroscopy qualitative analysis method, which have low precision and poor accuracy, and cannot be applied to complex systems with small samples, the present invention provides a method with high precision, good accuracy, and effective application to complex systems with small samples. A universal qualitative analysis method of near-infrared spectroscopy

Images