Raman spectrum multivariate data analysis method

Abstract

The invention relates to a Raman spectrum multivariate data analysis method. The method specifically comprises the following steps: collecting original Raman spectrum data of various samples by utilizing a Raman spectrum detection instrument; preprocessing the original data through an autonomous development analysis method; carrying out normalization and mean centralization data processing on the basis of spectral data preprocessing; aiming at the preprocessed data, selecting principal part analysis or least square discriminant analysis to extract spectral characteristic data, and respectively utilizing single factor variance analysis and cross validation to select significant characteristic parts in the spectral data; spectral recognition is carried out by combining the classification models, and the reliability of each classification model is evaluated by using unbiased preserving-one-method cross validation; and selecting residual samples for testing to obtain the accuracy, sensitivity and specificity of sample classification and the recognition performance of the tested worker characteristic curve evaluation system of the model. The method can be widely applied to the fields of Raman spectrum information processing and spectral characteristic computer recognition.

Description

technical field [0001] The invention relates to the field of Raman spectrum information processing and its spectrum feature computer identification, in particular to a Raman spectrum multivariate data analysis method. Background technique [0002] Raman spectroscopy is based on the interaction of light and chemical bonds within materials, and is a non-destructive analytical technique that can obtain detailed information on a sample's chemical structure, phase and morphology, crystallinity, and molecular interactions. Raman spectroscopy can also transfer the energy spectrum of molecules in the infrared region to the visible region for detection. Therefore, as a supplement to infrared spectroscopy, Raman spectroscopy is a powerful weapon for studying the structure of molecular substances. With the development and progress of science and technology, Raman spectroscopy has been applied in many fields such as petroleum, chemical industry, materials, biology, environmental protec...

AI Technical Summary

Problems solved by technology

First of all, Raman scattering is a weak optical phenomenon, and the spectral information (ie, Raman spectrum) generated by it is easily disturbed by the environment and external factors; secondly, in complex biochemical environments or other systems, different types of Biomacromolecules contain similar biochemical compositions, which lead to overlapping peak positions, uneven peak intensities, and extended peak widths (full width at half maximum) in their Raman spectra.

Images