偏最小二乘法在红外光谱识别茶叶中的应用

采用漫反射傅立叶变换红外光谱(FTIR)法结合主成分分析(PCA)、偏最小二乘法(PLS)、簇类的独立软模式(SIMCA)识别法对十三种茶叶进行了分类判别研究。研究结果表明,通过多元散射校正(MSC)对原始光谱进行预处理,可以提高模式识别技术的分类判别效果。在此基础上,选取1 900~900 cm-1波长范围内的茶叶红外光谱建立识别模型,三种方法都得到了满意的分类判别效果。在对检验集中全部130个样本的判别中,PCA仅有两类样本无法判别,SIMCA的识别率和拒绝率都在90%以上,而PLS的识别效果最佳,全部样本都得到了正确的归类。这一研究结果表明傅立叶变换红外光谱法与化学计量学方法相结合可以实现茶叶品种的快速鉴别,这为茶叶的客观评审提供了一种新思路。

Discrimination of Tea Varieties by Mid-Infrared Spectroscopy Combined with PLS

Diffusion reflectance FTIR spectroscopy combined with principal component analyses(PCA),partial least squares(PLS) and soft independent modeling of class analogies(SIMCA)respectively have been applied for classification analysis of the 13 tea varieties.Multiplicative scatter correct(MSC) spectral pretreatment was studied to improve the classification effect of pattern recognization models.The spectra between 1 900 cm-1 and 900 cm-1 were selected to set up the recognization models and most of tea varieties were discriminated correctly.In validation set of 130 tea IR spectra,just two varieties were mistaken with PCA,the recognition rates and rejection rates were both more than 90% with SIMCA,and the prediction correctly rates were 100% with PLS.This combination of FTIR spectroscopy and chemometrics was proved to be quick and reliable for tea identification,which would provide a new idea for tea authentication impersonally.