Combination of Excitation-Emission Matrix Fluorescence Spectroscopy and Chemometric Methods for the Rapid Identification of Cheaper Vegetable Oil Adulterated in Walnut Oil

Yan Zhang,Hai-Long Wu,An-Qi Chen,Ming-Yue Dong,Tong Wang,Xiao-Zhi Wang,Yu-Qin Yu
DOI: https://doi.org/10.1007/s11694-022-01536-6
IF: 3.4
2022-01-01
Journal of Food Measurement & Characterization
Abstract:Walnut oil (WNO) is considered as natural functional food of high economic interest due to its nutritional and medicinal benefits. Adulteration in WNO seriously damages the interests of consumers and the health of market. In this paper, a new strategy has been developed to identify the adulteration of WNO with other cheaper vegetable oils. The study was based on the chemometric analysis of excitation-emission matrix (EEM) fluorescence spectroscopy data of WNO samples containing different adulterants. First, alternating normalization-weighted error (ANWE) method was used for the data decomposition of oil samples to obtain spectral characteristics and chemical meaning information. Then, three pattern recognition methods were employed to build the classification models, including partial least squares discriminant analysis employing ANWE scores (ANWE-PLS-DA), k-nearest neighbor (kNN), and N-way partial least squares discriminant analysis (N-PLS-DA). Results showed that all models obtained good classification performances for the WNO and other vegetable oils (case 1). Moreover, N-PLS-DA outperformed ANWE-PLS-DA and kNN in the identification of pure and adulterated WNO samples (case 2). The accuracy rates of N-PLS-DA were 87.1–90.6
What problem does this paper attempt to address?