拉曼光谱
生物系统
多元分析
化学计量学
近红外光谱
偏最小二乘回归
模式识别(心理学)
主成分分析
灵敏度(控制系统)
分析化学(期刊)
红外光谱学
数学
认证(法律)
多元统计
人工智能
化学
统计
线性判别分析
红外线的
计算机科学
光学
色谱法
机器学习
生物
物理
工程类
有机化学
计算机安全
电子工程
作者
Ana M. Jiménez-Carvelo,Margherita Tonolini,Orla McAleer,Luis Cuadros-Rodrı́guez,Daniel Granato,Anastasios Koidis
标识
DOI:10.1016/j.foodres.2021.110196
摘要
Many different versions of vanilla extracts exist in the market in a variety of origins, purity levels and composition with little effective regulation. In this study, vanilla is authenticated both in terms of purity and geographical origin applying a multivariate approach using near infrared (NIR), mid infrared (MIR) and Raman spectroscopy following a complex experimental design. Partial least squares-discriminant analysis (PLS-DA) was applied to the spectral data to produce qualitative models. The prediction accuracy of the models was externally validated from the specific success/error contingencies. The results showed that MIR and Raman are reliable for authenticating vanilla in terms of purity, obtaining sensitivity, specificity, precision, and efficiency values equal to 1.00, and Raman is especially suitable for indicating the geographical origin of vanilla extracts, achieving performance metrics around 0.9.
科研通智能强力驱动
Strongly Powered by AbleSci AI