采集四川、 重庆、 云南、 贵州、 陕西五省市8个不同产地205个花椒样品的近红外光谱, 使用主成分分析(principal component analysis, PCA)、 判别偏最小二乘法(discriminant partial least squares, DPLS)分析了花椒产地的分类鉴别。 结果表明: 在12 500～3 800 cm-1波数范围内, 采用不同的光谱预处理方法可以建立较为稳健的DPLS模式识别模型, 对不同产地的花椒有较好的分类鉴别。 其校正集交叉验证除了经一阶微分预处理的模型识别率为99.39%外, 其他预处理方法识别率均为100%, 独立验证集总体识别正确率在85.37%~97.56%之间, 其中经标准正态变量变换(standard normal variate, SNV)、 多元散射校正(multiplicative scatter correction, MSC)预处理后的DPLS判别模型效果最好, 误判数仅分别为1个, 表示该方法在花椒产地识别中具有可行性。

Identification method of 205 Huajiao samples from 8 different geographical origins by near infrared spectroscopy coupled with principal component analysis (PCA) and pattern recognition based on discriminant partial least squares (DPLS) was proposed in this paper. In the spectra region between 12 500～3 800 cm-1, predictive models with different pretreatments of calibration set were built separately, and robust models indicating these geographic origins of Huajiao samples could be achieved using DPLS pattern recognition method. The correct identification rates of the independent validation set were between 85.37%~97.56%, in which DPLS discriminant model with standard normal variate (SNV) or multiplicative scatter correction (MSC) preprocessing was best. The method was effective in Huajiao origin recognition.