中文摘要：

采集了22个合格液态牛奶样品,并制备了50个掺入不同含量三聚氰胺（0.1～1500mg·kg^-1）的牛奶样品,应用近红外光谱仪扫描其透射光谱,研究利用近红外光谱快速检测牛奶中三聚氰胺的可行性。采用偏最小二乘法建立近红外光谱与牛奶中三聚氰胺含量之间的定量模型,结果表明近红外光谱受检测限的限制,难以准确预测牛奶中掺入的三聚氰胺的含量。而应用近红外光谱,结合判别偏最小二乘法建立定性模型,则可以实现对合格牛奶及掺入三聚氰胺的牛奶的定性鉴别,正确识别率达100%。因此,基于近红外光谱的检测方法可以初步判断牛奶中是否含有三聚氰胺,作为高效液相色谱法的补充,为定量检测做初步的筛查,可大大提高检测效率。

英文摘要：

In the present study,22 certified milk samples without melamine were collected,then 50 adulterated milk samples with added different content of melamine （0.1-1 500 mg·kg^-1） were prepared. The near-infrared （NIR） spectra of these milk samples were measured. The possibility of using NIR spectra to detect melamine in milk was studied. Partial least square regression （PLSR） was applied to construct the calibration model between NIR spectra and the content of melamine. The results showed that NIR spectroscopy can not accurately predict the content of melamine because of its poor detection limit. However,the combination of NIR spectra and partial least square-discriminate analysis （PLS-DA） was applied to differentiate the certified milk samples and the adulterated milk sample. The classification accuracy was 100%. Therefore,NIR spectra could be used to preliminarily detect whether the milk was adulterated with melamine. As a complementary detecting method to the high performance liquid chromatography （HPLC）,NIR spectra could improve the detecting efficiency of milk.