关键词: 山茶油, 拉曼光谱, 棕榈酸, 十四烷酸, 连续投影算法

Abstract:

Raman spectroscopy (RS) was applied for the rapid quantitative determination of fatty acids (palmitic acid and
tetradecanoic acid). Totally 70 camellia oil samples were prepared and subjected to Raman spectral analysis. Raw spectra,
wavelet derivative (WD) spectra and standard normal variate (SNV) spectra combined with successive projections algorithm
(SPA) were adopted to optimize partial least square (PLS) models for quantitative analysis of fatty acids. The results showed
that standard normal variate (SNV) and SPA could be used for the optimization of PLS models of fatty acids. The precision
of the models was improved, and modeling variables were simplified. The correlation coefficient of prediction (Rp) and root
mean square error of prediction (RMSEP) for the palmitic acid model were optimized from 0.917 and 2.191 to 0.981 and
1.078, and the modeling variables were reduced from 4000 to 44. For the tetradecanoic acid model, the Rp and RMSEP were
optimized from 0.940 and 0.058 to 0.969 and 0.047, and the modeling variables were 43. This study can provide a rapid
method for the analysis of fatty acids in camellia oil.

Key words: camellia oil, Raman spectroscopy, palmitic acid, tetradecanoic acid, successive projection algorithm