刘菁 高英 李倩

摘 要：针对乳制品缺乏全面系统的营养素成分检测数据，本研究应用成分检测与近红外（NIR）光谱建立了乳制品营养素快速检测模型。首先，通过台式和便携式两种NIR扫描仪获取乳样成分中营养素的NIR光谱。然后利用化学计量学软件选取适合的NIR片段，并通过NIR光谱导数、MSC、SNV等转换手段对选取片段进行了预处理。最后，建立了乳制品营养素的NIR光谱预测模型，确定了模型的参数，并通过实验验证了模型的有效性。结果表明，台式NIR光谱仪建立的模型优于微型NIR光谱仪建立的模型;马奶、牛奶、山羊奶的营养素通用模型预测精准度不高，有必要分别建立专用模型;PH对预测模型具有一定程度的影响。

关键词：乳制品成分检测;营养素;近红外光谱;NIR预测模型

中图分类号：TS252.7                         文献标识码：A                       文章编号：1001-5922（2021）07-0101-04

Nutrient Component detection and NIR Prediction Model Construction

Liu Jing， Gao Ying， Li Qian

（Institute of Clinical Nutrition， people s Hospital of Xinjiang Autonomous Region， Urumqi 830000， China）

Abstract：In view of the lack of comprehensive and systematic detection data of nutrients in dairy products， this study applied component detection and near-infrared （NIR） spectroscopy to establish a rapid nutrient detection model for dairy products. First，a rapid detection model was established by near infrared （NIR） spectroscopy. The NIR spectra of nutrients in milk samples were obtained by two NIR scanners， desktop and portable. Then the suitable NIR segments were selected by chemometrics software， and the selected fragments were pretreated by NIR spectral derivative， MSC， SNV and other conversion methods. Finally， the NIR spectrum prediction model of milk nutrients was established， and the parameters of the model were determined， and the effectiveness of the model was verified through experiments. The results showed that the model established by desktop NIR spectrometer was better than that established by micro NIR spectrometer; the accuracy of general nutrient model for horse milk， milk and goat milk was not high， so it was necessary to establish special models respectively; pH had certain influence on nutrient prediction model.

Key words：dairy ingredients detection; nutrients; near infrared spectroscopy; NIR prediction model

隨着国内生活水平的快速提高，人们对食物的营养成分要求越发严格，既要求营养均衡，同时又要求营养丰富。乳制品作为日常生活不可或缺的食物之一，其营养成分更是备受关注。近年来，针对乳制品鲜牛奶、酸牛奶的营养成分分析的检测研究成果较多，如娄文琦、罗汉鹏等人对牛奶的中红外光谱相关指标及遗传规律进行了研究[1]，剧柠、胡婕研究了光谱技术在乳及乳制品的应用进展[2]。虽然这些研究在一定程度上对牛奶等常见乳制品成分进行了有效检测，但对于马奶、山羊奶及其奶制品的成分研究较少。因此，为建立马奶、山羊奶及乳制品营养成分预测模型，本研究应用成分检测与NIR预测模型，构建了马奶营养素NIR预测模型，重点对乳制品中的营养素含量进行了预测。……