YAO shule, HU Yufeng, SONG Xiaodong, ZHENG Nan, WANG Jiaqi, ZHAO Shengguo
China Dairy. 2025, (11): 58-65.
[Objective] This study explored using miniaturized near-infrared (NIR) spectroscopy to detect Holstein milk adulteration in specialty milks. [Method] Spectral data of camel,yak,buffalo,goat milk,and their adulterated samples with Holstein milk were collected via miniaturized transmissive and reflective NIR spectrometers. Then,quantitative models were built with partial least squares regression,while qualitative models were developed using support vector machine (SVM) and linear discriminant analysis (LDA). [Result] Quantitative models showed good predictive ability. Transmissive predictive models achieved a coefficient of determination(R²) of 0.78~0.99 and a root mean square error (RMSE) of 2.31~13.68,while reflective ones got 0.89~0.98 and 4.68~9.77,respectively. In qualitative analysis,SVM outperformed LDA in classification. Transmissive-spectrum SVM reached 90.91%~100.00% accuracy,and reflective-spectrum SVM 80.00%~100.00%. [Conclusion] This study indicated that miniaturized NIR spectrometers,as fast and convenient detection tools,hold great application potential. The devices enable on-site and real-time detection of milk adulteration,offering strong technical support for dairy quality control and market regulation.