| 摘要: |
| 【目的】建立牛奶中尿素氮的快速、无损检测方法，为牛奶中尿素氮的快速检测提供支持。【方法】对200个牛奶样品进行近红外扫描，并用多功能乳制品分析仪对牛奶样品中尿素氮的含量进行测定；剔除20个异常样品后，得到由180个牛奶样品组成的得分样品，将得分样品分为定标集（144个）和验证集（36个），将正交试验设计与主成分回归法（PCR）、偏最小二乘法（PLS）、改进偏最小二乘法（MPLS）3种定量校正方法和多种光谱预处理方法结合，建立牛奶中尿素氮的近红外检测模型，利用目标函数法对模型预测效果进行评定。【结果】建立了定量检测牛奶中尿素氮的最优模型，其定标相关系数（R2）和定标标准差（SEC）分别为0.986 4和0.238 4。用验证集对所建模型进行验证，其校正相关系数（RSQ）和预测标准差（SEP）分别为0.976 0和0.360 0。利用所有得分样品对预测结果进行监控，并绘制尿素氮测定值与模型预测值的线性相关曲线，相关系数r2为0.980 5。【结论】利用近红外光谱法建立的尿素氮定量检测最优模型具有很好的适用性和准确性，可用于牛奶尿素氮的快速定量检测。 |
| 关键词: 牛奶品质 近红外光谱 尿素氮 快速检测 |
| DOI： |
| 分类号: |
| 基金项目:陕西省青年科技新星项目（2014KJXX-51）；中国博士后科学基金项目（2015M570811）；中央高校基本科研业务费专项（GK201505126）；国家科技支撑计划项目（2012BAD12B07） |
|
| Quantitative detection of milk urea nitrogen using near infrared spectroscopy |
|
LIU Yongfeng,LI Shuanghong,KU Ting,et al
|
| Abstract: |
| 【Objective】A method that can quickly and non-destructively measure milk urea nitrogen was established to provide support for detection of urea nitrogen in milk.【Method】A total of 200 milk samples were scanned by InfraXact NIR analyzer,and the contents of urea nitrogen were measured using the multifunctional analyzer of dairy.After eliminating 20 abnormal samples,the scored 180 milk samples were grouped to the calibration set (144 samples) and the validation set (36 samples).Then,the orthogonal design was combined with principle component regression (PCR),partial least square (PLS),modified PLS and multiple pretreatment methods to establish the model of milk urea nitrogen.The objective function method was used to evaluate the model.【Result】The optimal model for quantitatively testing milk urea nitrogen was built with the calibration coefficient (R2) and standard error of calibration (SEC) of 0.986 4 and 0.238 4.The correlation coefficient (RSQ) and standard error of prediction (SEP) when using validation set to examine the model were 0.976 0 and 0.360 0.Finally,all scored samples were used to monitor the predicted results and the linear correlation curve between the measured and predicted values was drafted,which had the correlation coefficient r2 of 0.980 5.【Conclusion】The optimal model for quantitatively testing milk urea nitrogen based on near infrared spectroscopy had good applicability and accuracy. |
| Key words: milk quality near infrared spectroscopy milk urea nitrogen rapid detection |
