| At present,people pay more and more attention to the quality of traditional Chinese medicine and its preparations.With the efforts of researchers and the development of analytical techniques,the quality control of traditional Chinese medicine is gradually developing in a comprehensive and integrated direction.In the market,some vendors have arbitrarily marked the origin of codonopsis,and the adulteration and fraud of codonopsis has become more and more serious,which has seriously affected the normal market order.Therefore,comprehensive,rapid and effective quality control of the quality of traditional Chinese medicine and its preparations is an important task at present.When identifying the origin and quality of codonopsis pilosula,the disadvantages of traditional morphological identification methods are that they must have sufficient experience and certain subjective factors;the shortcomings when using chemical methods to identify the origin and quality of codonopsis pilosula are manifested in the detection time.Long,complicated operation steps,and a lot of organic reagents are needed.Therefore,it is particularly important to establish a fast and concise method of identification and quality analysis of Codonopsis pilosula origin.Because hyperspectral technology has the advantages of non-invasive,non-destructive,and non-polluting.Therefore,this article uses hyperspectral technology combined with chemometric methods to identify and analyze the origin of codonopsis;at the same time,quickly determine the polysaccharides of codonopsis to achieve rapid,non-destructive and comprehensive quality control of codonopsis.The main findings are as follows:1)This study collected the hyperspectral data of 26 samples of codonopsis pilosula,extracted the effective information in the spectrum by chemometric methods,and identified the origin of codonopsis pilosula from different places.The results showed that under the pretreatment of multivariate scattering correction(MSC),principal component analysis(PCA)could basically classify codonopsis pilosula samples from different origins in the range of 350~2500nm.Then,the random forest algorithm was used to identify the origin of samples from Changzhi and non-Changzhi and Gansu and Changzhi.The correct rates of identification of the model were99.7% and 66.7%,respectively,indicating that the use of hyperspectral technology could quickly and easily achieve the identification of codonopsis Identification of origin.2)This study performed multivariate scattering correction(MSC),logarithm(LG),first-order differential(1st),second-order differential(2nd),standard normal transformation(SNV),1/R,and SG transformation on the original spectrum,and studied the best spectral pretreatment method of Codonopsis pilosula polysaccharide;and use continuous projection algorithm(SPA)and load method to extract and determine the important band of codonopsis polysaccharide,and finally constructed a estimation of codonopsis pilosula linear model based on partial least squares method and multiple linear regression method.with nonlinear models(BP-ANN,support vector machine regression).The main conclusions are as follows:Different spectral pretreatment methods could improve the correlation between some spectral bands and polysaccharides to vary degrees.Among them,the 2nd transform spectrum had the highest correlation with and the correlation coefficient could reach up to 0.731.Different types of models have different monitoring effects on Codonopsis pilosula polysaccharides.The best monitoring model was based on the multiple linear regression model under SNV 2nd transform spectrum(correction set Rc2=0.923,RMSEC=0.028,RPD=3.321 validation set Rp2=0.836,RMSEP= 0.051,RPD=1.156).In the extraction of characteristic bands,the continuous projection algorithm is better than the load method,which showed that the quantitative analysis of codonopsis pilosula using hyperspectral analysis technology was feasible. |
PDF Full Text Request