采用了近红外光谱法测定了4个品牌牙膏样品的近红外光谱, 然后对所得的光谱进行变量预处理, 再运用偏最小二乘法(PLS)、 人工神经网络法(ANN)和K-最近邻法(KNN)等几种有监督模式识别法, 以及主成分分析(PCA)和聚类分析两种无监督模式识别法对样品进行了品牌的分类及聚类分析。 结果表明采用近红外光谱法所得的光谱变量经多元散射处理后的分类结果比较理想。 各品牌牙膏的质量相对是稳定, 但在采集的四种品牌牙膏中, 有两种品牌的牙膏样品间存在较大差异。

The near-infrared spectroscopy (NIR) was combined with pattern recognitions method and applied to the quality assessment of toothpaste samples of four different brands. Several chemometrics approaches, such as principal component analysis (PCA), clustering analysis (CA), partial least squares (PLS), artificial neural networks (ANN) and K-nearest neighbor (kNN) were used to investigate the quality of toothpastes samples. The obtained results showed that the four clustering groups can be observed after the pretreatment of multiple scatter correction for the NIR data. It was also found that the quality of toothpastes of all the four brands was relatively stable, however, there is a significant difference in the quality between two brand kinds of toothpaste samples.