近红外光谱技术对液态奶制品中乳糖、脂肪及蛋白质的定量检测

用InfraXact多功能近红外分析仪和正交实验设计，研究了不同预处理和校正方法对建模效果的影响，建立了液态奶制嘉中脂肪、乳糖和蛋白质质量含量的定标模型。利用目标函数法对模型进行评定，并对选出的最优模型的适用性进行验证．结果表明，最优模型对糖、脂肪和蛋白质的相对标准偏差分别为4．12％，7．04％，9．61％，相对误差分别为2．82，6．28，2．72．正交实验方法能综合考虑不同预处理方法和校正方法对模型建立的影响．近红外光谱分析法能够实现对牛奶中脂肪、乳糖和蛋白质质量含量进行快速、无损的定量检测．

Near-infrared Spectroscopy on the Quantitative Detection of Lactose, Fat and Protein in Milk Products

Using InfraXact type multi-functional near-infrared analyzer, the effect of different pretreatment and calibration methods on modeling were studied through orthogonal test design. The calibration model for lactose, fat and protein content in liquid milk products were established. The models were evaluated by objective function method and the applicability of the optimal models were also validated. The results showed that relative standard deviation of the three optimal models were 4.12%, 7.04% and 9.61%, and the ratio of performance to standard deviation were 2.82, 6.28 and 2.72, respectively. Research results showed that orthogonal experimental design could evaluate the impact of different pretreatment and calibration methods to building an optimal modeling. The lactose, fat and protein content of liquid milk products can be rapidly and nondestructively detected by near infrared spectroscopy in a quantitative way.