Abstract

Spatial heterodyne Raman spectroscopy has been widely applied in various fields due to its non-contact, nondestructive, fast, high stability, and high spectral resolution characteristics. This article integrated spatial heterodyne Raman spectroscopy with chemometric methodologies to assess the feasibility of peak-to-peak ratio regression, partial least squares regression, support vector machine regression, and non-negative matrix factorization for the quantitative analysis of mixtures. Chemometrics methods were used to model and analyze the interference data in the interferogram domain, and variational mode decomposition was used to extract features from the interference data, further improving the interference data’s modeling and prediction accuracy. The results demonstrate that modal intensities obtained through variational mode decomposition of interference data improve the prediction accuracy of regression analysis. Support vector regression exhibited the most favorable predictive performance among the tested models. The root mean square error of cross-validation was reduced from 5.55% to 2.64%, and the prediction root mean square error was reduced from 5.08% to 1.5%. The improved model utilizing interference data showed higher fitting accuracy and more precise sample predictions compared to spectral data modeling.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call