近红外光谱法判别鲍鱼品种的研究

Classification of AbaloneUsing Near Infrared NIR Spectroscopy

利用近红外光谱分析技术结合化学计量学方法对三种鲍鱼快速分类方法进行研究,为鲍鱼分类提供一种新的方法和手段。使用JDSU便携式红外光谱仪采集产地为莆田的91只,18月龄的绿盘鲍、皱纹盘鲍和红壳鲍三种鲍鱼的光谱数据。分别采用簇类独立软模式法(SIMCA)和偏最小二乘判别分析法(PLSDA)建立三种鲍鱼的快速分类模型。结果表明,一阶导数处理后SIMCA模型达到最优;先平滑后二阶导数或者单独使用均值中心化处理能使PLS-DA模型达到最优,其中PLS-DA模型效果更好,训练集和检验集的分类识别正确率分别为100%和97%,能满足鲍鱼现场快速分类的要求。

We combined the NIR spectroscopy and chemometrics methods to build abalone classification models.The NIR spectra were collected with JDSU micro NIR spectrometer.The 91 samples were Green Abalone,Pacific Abalone and Red Abalone,which were 18 months old and came from Putian.The PLS-Discriminant Analysis(PLS-DA)and soft independent modeling of class analogy(SIMCA)were used to build three-class abalone classification models.The results show thatthe accuracy of training setand validation set of PLS-DA respectivelyare 100% and 97%.