Detection Of The Adulteration Of Panax Notoginseng Powder Based On Terahertz Spectroscopy

Panax Notoginseng is an important Chinese medicinal material in China.It is well known for its functions of hemostasis,anti-hypertension,anti-thrombosis and nerve protection in the world.The price of Panax Notoginseng is very expensive because of the limited production.In the market,some businesses in order to profiteers,the phenomenon of Panax Notoginseng powder adulteration is very serious.It greatly affects the quality and safety of Notoginseng powder and the physical and mental health of consumers.Therefore,it is of great significance to study the quality inspection of Panax Notoginseng powder.In this paper,terahertz time domain spectroscopy(THZ-TDS)combined with Chemometrics was used to detect several common adulterations of Panax Notoginseng powder.The qualitative and quantitative analysis model was established by using the algorithm to achieve the purpose of identification of adulterated Notoginseng powder.The research contents and conclusions are as follows:(1)Realize the rapid identification of four different underground parts of Panax Notoginseng based on terahertz spectroscopy,and establish a qualitative analysis model.In this paper,the Main roots,Rhizome,Branch root And Fibrous roots of Panax Notoginseng different parts were studied.A total of 133 samples were prepared and 532 terahertz spectra were obtained.Firstly,the refractive index,absorption coefficient,time-domain spectrum and frequency-domain spectrum of the sample were analyzed,and it was found that the terahertz spectrum of different parts of the samples was significantly different.Then the absorption coefficient spectrum was used to establish the model.Second,pretreatment algorithms such as Baseline Correction(BC),Multiplicative Scatter Correction(MSC)and Normalization were used to preprocess the absorption coefficient spectrum of 0.5-2.0 THz,to remove noise and improve the spectral signal-to-noise ratio.Then,use the KS algorithm to divide the modeling set and the prediction set at a ratio of 3:1 by class,and then use the Successive Projections Algorithm(SPA)to select the characteristic frequency points of the sample,and finally input the selected characteristic variables into the support vector machine(SVM)And Linear Discriminant Analysis(LDA)algorithm,establish a qualitative analysis model.The results show that the SPA-SVM model under the linear kernel function has the best results after the Baseline Correction pretreatment.The training accuracy of the model is 99.75%,and the accuracy of the test set is 99.25%.In the SPA-LDA model,the result of Baseline Correction pretreatment is the best.The training accuracy of the model is 100%,and the accuracy of the test set is 100%.The results show that terahertz spectroscopy can be used for the identification of different parts of Panax Notoginseng,which provides a fast and feasible method for the detection of different parts of Panax Notoginseng.(2)Terahertz spectroscopy was used to identify the mixture of Panax Notoginseng powder and Rhizome powder.Taking Notoginseng powder(1:0),Rhizome powder(0:1)and the mixture of Main roots and Rhizome powder(2:1,1:1)as the research objects,Firstly,the spectra of the samples were analyzed.Then,the absorption coefficient spectra and refractive index spectra of the four proportion samples were analyzed by using principal component analysis(PCA).The results show that the four types of samples have obvious clustering,and the clustering effect of refractive index spectrum is better than that of absorption coefficient spectrum.Subsequently,the refractive index spectrum and its first three principal components were respectively imported into the partial least square discriminant analysis(PLS-DA)and principal component regression discriminant analysis(PCR-DA)algorithm to establish the model.The results showed that the PLS-DA model could separate 100%Panax Notoginseng powder,Xiankou powder and their mixture.But mixtures are hard to separate.The results showed that terahertz spectrum could be used to determine whether the powder contained in the Notoginseng powder and whether the powder was removed in the taproot of Notoginseng powder during processing.This is of great significance to the quality evaluation of Panax Notoginseng powder.(3)The terahertz spectroscopy was used to quantitatively and qualitatively analyze the adulteration of similar foreign substances in Panax Notoginseng powder.In this experiment,four types of samples were prepared,three types of adulterated samples,Notoginseng powder adulterated zedoary turmeric powder,Notoginseng powder adulterated wheat flour and Notoginseng powder adulterated rice flour,the adulteration concentration was 5%～60%.The concentration interval is 5%,the other type was pure Notoginseng powder sample.Through spectral comparison,it is found that the spectra of different kinds of adulterated samples and different concentrations of the same kind of adulterated samples have obvious differences.firstly,Qualitative analysis of three types of adulterated substances.The SPXY algorithm was used to divide the modeling set and the prediction set 3:1,Then UVE and SPA were used to select the feature information.Finally,the selected feature information was imported into BP neural network(BPNN)to establish the qualitative analysis model of BPNN.The model results showed that the classification accuracy of the UVE-BPNN qualitative model was 95%,and that of the SPA-BPNN qualitative analysis model was 92.25%.Then,the quantitative analysis models of least squares support vector machine(LS-SVM)and partial least squares discriminant analysis(PLS)were established by terahertz absorption coefficient spectrum.The modeling set and prediction set were randomly divided by class in a 3:1 ratio.Then,the least-squares support vector machine(LS-SVM)and partial least squares(PLS)were used to establish the quantitative analysis model.The results showed that the quantitative model established by LS-SVM for the identification of the content of zedoary turmeric powder and wheat flour was better,and the correlation coefficient(R_P)of the prediction set was 0.9015 and0.9305,and the root mean square error(RMSEP)of the prediction set was 0.0723 and 0.0677,respectively.The quantitative model established by PLS for the identification of rice flour adulteration is better,which R_P is 0.9424 and RMSEP is 0.0601.The quantitative model has achieved a good prediction effect,the correlation is greater than 0.9,and the error is controlled in a relatively small range.The results showed that terahertz spectroscopy could be used for the qualitative and quantitative detection of similar substances(zedoary turmeric powder,wheat flour and rice flour)in Panax Notoginseng powder.This study provided a research basis for the detection of terahertz adulteration of similar substances in Panax Notoginseng powder.