Quantitative analysis of alum based on Terahertz time-domain spectroscopy technology and Support vector machine

Abstract

In this paper, the Terahertz time-domain spectroscopy (THz-TDS) technique was used to quantitatively analyze the alum proportion in sweet potato starch. Terahertz time-domain spectra and frequency-domain spectra of pure sweet potato starch and pure alum and their mixtures were obtained. The absorption coefficient spectrum and refractive index spectrum of the samples were calculated. Results show that alum exhibit unique absorption peaks in the terahertz range, and can thus be identified with its absorption fingerprints in the detection from Sweet potato starch. Differences in the refractive index can also be observed between the alum and the sweet potato starch. The Partial Least Squares (PLS) and the Support vector machine (SVM) algorithm were used to establish the mathematical model between the absorption coefficient and the proportion. Results show that the prediction accuracy can reach 99.8% and 99.9%. Comparing the two predictive models, SVM has a better prediction effect. In all, the THz-TDS system plus SVM method is very promising for the further quantitative analysis of alum in sweet potato starch, with the characteristics of being nondestructive and laborsaving compared to other analytical tools.