近红外光谱和化学计量学对不同产地不同产期当归的定性研究

采用近红外光谱技术结合化学计量学,对来源于不同产地不同产期的136个当归样本进行快速、无损的定性研究.将当归样本粉碎后,采用光纤漫反射采集近红外光谱,并对光谱进行主成分分析(PCA)、结合随机森林建立当归的定性模型,同时对比分析了不同光谱预处理方法(未处理、变量标准化十一阶导数、变量标准化十二阶导数)对鉴别结果的影响....

Qualitative Research on Radix Angelicae Sinensis from Different Origins in Different Harvest Time by Near Infrared Spectrum and Chemometrics

The 136 Radix Angelicae sinensis samples from different origins in different harvest time were fast,nondestructively and qualitatively detected by near infrared（NIR） spectrum technique with chemometrics.After the samples were crashed,the NIR spectrum were collected by fiber-optic in diffuse reflectance mode,and the spectrum were analyzed by principal component analysis（PCA）,and the qualitative model of Radix Angelicae sinensis were established combining with random forests（RF）,meanwhile,the effects of the different pretreatment methods,such as untreated,SNV ＋ first derivative and SNV ＋ second derivative,on identification cause were analyzed.The results of PCA and RF showed that the samples could separate at product time and origins,and Radix Angelicae sinensis were closed in the same product time,and the classification accuracy of random forests reached 94.85% with treatment by SNV ＋ first derivative.The principle of NIR spectrum combining with RF was simple with simplicity of operation and high accuracy,and the origins and product time of Radix Angelicae sinensis were synchronously identified.