Identification of Radix puerariae starch from different geographical origins by FT-NIR spectroscopy. (29th December 2017)

• Record Type: Journal Article
• Authors: Zhang, Hang; Jiang, Hui; Liu, Guohai; Mei, Congli; Huang, Yonghong
• Is Part Of: International journal of food properties. Volume 20(2017)Supplement 2
• Page Start: 1567
• Page End: 1577
• Publication Date: 2017-12-29
• Subjects: ELM -- FT-NIR spectroscopy -- IRIV -- Radix puerariae starch -- siPLS; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664.0705
• DOI: 10.1080/10942912.2017.1283325
• Languages: English
• ISSNs: 1094-2912

Abstract

ABSTRACT: Fourier transform near-infrared (FT-NIR) spectroscopy technique combined with multivariate calibration approach was employed to identify the geographical origins of Radix puerariae starch. First, the efficient spectral subintervals were selected by a synergy interval partial least squares (siPLS) method. Second, an iteratively retains informative variables (IRIV) algorithm was applied to select the characteristic wavelengths from the efficient spectral subintervals obtained by siPLS. Experimental results showed that the number of wavelength variables obtained by IRIV was 10. Meantime, only the first two PCs of principal component analysis (PCA) based on the selected 10 variables could explain 99.9769% of the total variance and the identification rate of the validation set is 100% based on extreme learning machine (ELM) in this study. This work indicates that the FT-NIR spectroscopy analysis technique coupled with multivariate calibration is an excellent choice for the discrimination of geographical origins of Radix puerariae starch. In addition, the IRIV is a promising algorithm for the selection of characteristic wavelength variables in the practical application.