近红外光谱结合siPLS法用于深度水解蛋白奶粉掺伪的快速检测

NIR Spectroscopy Combined with siPLS Method for Rapid Detection in Enzymatically Hydrolyzed Protein Milk Powder′s Adulteration

目的:建立深度水解蛋白奶粉中掺伪普通蛋白粉的快速检测方法。方法:向深度水解蛋白奶粉掺伪一定比例的牛乳清蛋白粉和植物蛋白粉,共制备171个掺伪样品,并采集近红外光谱;对采集的样品光谱使用SPXY法按3∶1比例划分为校正集和预测集,应用联合区间偏最小二乘法(siPLS)建立掺伪检测模型,并比较不同预处理方法下的建模效果。结果:SG一阶导预处理下建立的siPLS模型效果最好,其组合区间光谱范围为[1135~1239.5,1660~1764.5,2080~2184.5 nm],校正集相关系数R^(2)为0.9948,RMSECV值为0.0101,预测集相关系数R^(2)为0.9945,RMSEP值为0.0110,RPD值为13.5。结论:通过siPLS法筛选光谱区间建模,可提高模型的稳定性和预测精度,本方法操作简便,可用于深度水解蛋白奶粉中的掺伪蛋白粉的快速无损检测。

Objective:This study aims to develop a rapid detection method for the adulteration of common protein powder in enzymatically hydrolyzed protein milk powder.Methods:The whey protein powder and plant protein powder were mixed in specific proportions into enzymatically hydrolyzed protein milk powder,resulting in 171 adulterated samples.NIR spectra were then collected.The obtained spectral data were divided into calibration and prediction sets using the SPXY method in a 3∶1 ratio.The adulteration detection model was established using the siPLS method and preprocessed with various methods.Results:The siPLS model established through SG first-order derivative preprocessing exhibited the best performance.The combined interval spectral range was[1135 to 1239.5,1660 to 1764.5,2080 to 2184.5 nm].The calibration model had a correlation coefficient(R^(2))of 0.9948,with an RMSECV value of 0.0101,while the prediction model had an R^(2) of 0.9945,with an RMSEP value of 0.0110,and an RPD of 13.5.Conclusion:Modeling with siPLS method can improve the stability and prediction accuracy.The method is simple to operate and can be applied for the rapid non-destructive detection of adulterated protein powder in enzymatically hydrolyzed protein milk powder.