化学计量学
偏最小二乘回归
主成分分析
质子核磁共振
化学
牛乳
食品科学
代谢组学
掺假者
核磁共振波谱
色谱法
计算机科学
机器学习
人工智能
立体化学
有机化学
作者
Jing Cui,Dan Zhu,Meicheng Su,Dongfei Tan,Xia Zhang,Man Jia,Gang Chen
摘要
The illegal undeclared addition of reconstituted milk powder to ultra-heat treated (UHT) milk to lower production costs is an example of economically motivated adulteration. This activity not only defrauds consumers but also places honest traders at a disadvantage, which could damage the reputation of milk producers and reduce the integrity of the markets. In this research, a non-targeted analytical strategy that combines proton (1 H) nuclear magnetic resonance (NMR) spectroscopy with a chemometrics data mining tool was developed for the authentication of bovine UHT milk.Unsupervised principal component analysis was used to distinguish UHT and tap-water-reconstituted powdered milk. Partial least squares-discriminant analysis (PLS-DA) with R2 (Y) and Q2 equal to 0.859 and 0.748, respectively, was used to differentiate UHT and reconstituted milk samples. Three compounds were selected as biomarkers to distinguish UHT and reconstituted milk and identified according to the standard NMR-spectra database. Finally, a PLS-DA model was established, according to the characteristic spectral bands, to identify UHT milk and reconstituted milk.This procedure demonstrated the feasibility of using non-targeted NMR profiling combined with chemometric analysis to combat mislabeling and fraudulent practices in milk production. © 2019 Society of Chemical Industry.
科研通智能强力驱动
Strongly Powered by AbleSci AI