Abstract Goat milk has a potential target of fraud. In this sense, Near Infrared Spectroscopy (NIRS) have been successfully used to detect food fraud. This study aimed to develop multivariate classification models using NIRS to detect adulterants in goat milk. Principal Component Analysis (PCA), control chart, k-Nearest Neighbor (k-NN), Part Least Square-Discriminant Analysis (PLS-DA) and Soft Independent Modeling of Class Analogies (SIMCA) were used to detect the adulterants: water, urea, bovine whey and cow's milk in goat's milk samples with concentrations of 0 (control), 1, 5, 10, 15 and 20% v/v, resulting in 300 control samples and 300 adulterated samples. The control chart discriminated authentic and adulterated samples with 95% confidence. The PLS-DA results were better compared to those obtained by k-NN and SIMCA; presenting 100% sensitivity and specificity in calibration, cross validation, and prediction. Therefore, NIRS combined with PLS-DA was adequate to detect goat milk safety control associated with adulteration.

中文翻译：

---

振动光谱和化学计量学工具的真实性和改进山羊奶的安全控制

摘要 山羊奶具有潜在的欺诈目标。从这个意义上说，近红外光谱 (NIRS) 已成功用于检测食品欺诈。本研究旨在开发多变量分类模型，使用 NIRS 检测山羊奶中的掺假物。使用主成分分析 (PCA)、控制图、k-最近邻 (k-NN)、部分最小二乘判别分析 (PLS-DA) 和类类比的软独立建模 (SIMCA) 来检测掺杂物：水、山羊奶样品中的尿素、牛乳清和牛奶，浓度为 0（对照）、1、5、10、15 和 20% v/v，得到 300 个对照样品和 300 个掺假样品。控制图以 95% 的置信度区分了真实样品和掺假样品。与 k-NN 和 SIMCA 获得的结果相比，PLS-DA 结果更好；在校准、交叉验证和预测方面呈现 100% 的灵敏度和特异性。因此，NIRS 结合 PLS-DA 足以检测与掺假相关的山羊奶安全控制。