目的 利用近紅外光譜技術(shù)（near infrared spectroscopy，NIRS）建立三七藥材水分和醇溶性浸出物定量分析的快速測(cè)定方法。方法 參照《中國藥典》2015年版三七水分和醇溶性浸出物含量測(cè)定方法對(duì)53批藥材分別測(cè)定水分和醇溶性浸出物含量，采用偏最小二乘法（PLS）分別建立水分和醇溶性浸出物的近紅外定量分析模型，并利用內(nèi)部交叉驗(yàn)證和外部驗(yàn)證的方法對(duì)模型進(jìn)行優(yōu)化。結(jié)果 藥材樣品中水分和醇溶性浸出物預(yù)測(cè)最佳波段分別為4 450.32～7 350.01 cm^-1和6 163.92～3 984.71 cm^-1。定量模型校正集相關(guān)系數(shù)分別為0.997 2和0.962 4，校正均方差分別為0.039 6和0.776 0；驗(yàn)證集的相關(guān)系數(shù)為0.962 4和0.988 4，驗(yàn)證均方差分別為0.173和0.595。結(jié)論 該方法準(zhǔn)確、快速、無污染，可用于三七藥材中水分和醇溶性浸出物含量的快速測(cè)定。

Objective This study aimed at establishing the quantitative model of moisture and ethanol extract content in Panax notoginseng by NIRS. Methods Oven drying method of Chinese Pharmacopoeia 2015 was used to determine the moisture and ethanol extract content of 53 samples. The correlation models of the NIR spectra and the reference values of moisture and ethanol extract content were obtained with partial least square (PLS) method and optimized through inner cross validation and external validation. Results The optimal spectrum range of moisture and ethanol extract were 4 450.32-7 350.01 cm^-1 and 6 163.92-3 984.71 cm^-1. The correlation coefficients of calibration for moisture and ethanol extract were 0.997 2 and 0.962 4,and the root mean square errors of calibration were 0.039 6 and 0.776 0. The correlation coefficients of predication were 0.962 4 and 0.988 4, and the root mean square errors of prediction were 0.173 and 0.595. Conclusion The presented method is accuracy, timesaving and nondestructive, which could be applied for the rapid detection of the moisture and ethanol extract content in Panax notoginseng.

科學(xué)技術(shù)部國家重大新藥創(chuàng)制項(xiàng)目（2013ZX09402202）