中药饮片近红外光谱水分定量模型的研究

目的: 探索建立中药饮片近红外光谱水分定量模型的可行性和思路。 方法: 分别以5种饮片(全草或叶)设计的梯度水分样品和52种中药饮片为分析对象,采用中国药典干燥失重测定法测定水分含量,用光纤探头测定近红外漫反射光谱,应用OPUS软件处理,建立模型,比较分析模型参数。 结果: 5种饮片单独建立的水分定量模型,内部交叉验证决定系数均大于98.0,交叉验证均方差(RMSECV)均小于1.0;将5种饮片中的2种、3种、4种或5种分别混合建模,内部交叉验证决定系数均大于90.0,交叉验证均方差(RMSECV)均小于2.0。52种饮片建立的水分定量模型,内部交叉验证决定系数较低,交叉验证均方差(RMSECV)较大,验证结果不佳。 结论: 同一种饮片能够建立水分定量模型;同种类或不同种类但光谱相似的中药饮片建立模型可行,但预处理方法和测样条件等还有待进一步优化;不同种类的中药饮片在不做任何分类的情况下,不能建立通用性水分定量模型。

Objective: To explore the feasibility and ideas of the establishment of quantitative models of water content for Chinese herbal pieces by NIR. Methods: The samples of gradient water content designed with 5 kinds of herbal pieces(the full grass or leaves)and 52 kinds of Chinese herbal pieces were taken as the subjectives of this analysis.The water content was determined by the weight loss on drying method according to Chinese Pharmacopoeia.The near-infrared diffuse reflectance spectra of 5 kinds of samples(the full grass or leaves)with gradient water content and 52 different kinds of Chinese herbal pieces were determined by a fiber optic probe.The spectra were processed by OPUS software,the models were established,and the model parameters were compared and analyzed. Results: For the separate quantitative model of 5 kinds of Chinese herbal pieces,the internal cross–validation determination coefficients R² were higher than 98.0,the room mean square errors of cross–validation(RMSECVs)were less than 1.0.For the quantitative model of combination of two,three,four or five kinds of the above Chinese herbal pieces,the R² were higher than 90.0,the RMSECVs were less than 2.0.For the quantitative model of 52 different kinds Chinese herbal pieces,the R² were much smaller,the RMSECVs were larger,and the verification results were poor. Conclusion: Establishment of quantitative models of water content for each single kind of Chinese herbal pieces was feasible.Establishment of quantitative models of water content for the same kind or different kinds of Chinese herbal pieces with similar spectra was feasible too,but the pretreatment and determination methods should be further optimized.The universal quantitative models of water content for different kinds of Chinese herbal pieces could not be established without any classification.