Research On Quantitative Analysis Of The Near Infrared Spectra Of Gentiana Macrophylla

Gentiana is a precious traditional Chinese medicinal material in my country.It is a perennial herb of Gentianaceae.It has good curative effects in anti-inflammatory,analgesic and rheumatic arthralgia.It can be used as a medicinal plant with great development and utilization value.At present,wild Gentiana resources are in an endangered state.There are a lot of false ones and substandard ones in the market,which can easily cause confusion in medicine.The quality of Gentiana is not only related to the quality control of traditional Chinese medicine,but also related to the effect of clinical treatment.However,due to the complex chemical composition and ambiguous mechanism of action,the traditional method of quality detection of Gentiana is time-consuming and labor-intensive and consumes a lot of medicinal materials,which is difficult to meet the large-scale real-time detection of Gentiana and its drugs.In the process of material analysis,near-infrared spectroscopy has been widely used in the field of rapid detection of the quality of traditional Chinese medicine due to its advantages of non-destructive detection,simple operation,and high efficiency.In this study,the contents of gentiopicroside and loganic acid in Gentiana macrophylla were analyzed by near infrared spectroscopy,and the differences of different chemometrics methods were studied.In order to further improve the stability and prediction ability of the model,according to the characteristics of high dimensional small samples and sparse prior signals of near infrared spectral signals,a quantitative analysis model of OMP algorithm was proposed based on compressed sensing theory.It is possible to determine the quality of Gentiana macrophylla quickly.The main research contents and results are as follows:(1)Aiming at the problem of some irrelevant information interference in the collection process of near-infrared spectra,the influence of different spectral preprocessing methods on the model prediction results is studied.The collected 200 near-infrared spectral data of Gentiana were divided into training set and prediction set according to the KS method,and a variety of preprocessing methods were selected to establish partial least squares regression analysis model.The results show that different preprocessing methods have varying degrees of influence on the prediction results of the model,and the model after WT processing has a better prediction effect.(2)On the basis of WT as the preprocessing method,the input variables of the model are further optimized,and the Daubechies(db N)wavelet base with orthogonality and discrete wavelet transform function is selected to denoise the spectral signal.Taking the content of gentiopicroside and loganinic acid in Gentiana as the analysis object,the db5 wavelet basis function is selected,the number of decomposition layers is 6,and the wavelet threshold is set to 0.05 and 0.005 respectively,the RMSEC of the PLSR analysis model are 0.2468 and0.2565,respectively.R~2were 0.9470,0.9001,RMSEP were 0.2303,0.2499,R~2were 0.9187,0.8893.The results showed that the optimized wavelet transform was used as a pretreatment method for the content of gentian bitter glucoside and strychnine in Gentiana macrophylla,and the quantitative analysis model was well fitted and the prediction results were more accurate.(3)Aiming at the impact of the high-dimensional,time-varying,nonlinear,variable redundancy and other characteristics of the near-infrared spectrum on the quantitative analysis model,the compressed sensing technology is applied to the near-infrared spectrum technology,and a near-infrared spectrum based on the OMP algorithm is proposed.Quantitative Analysis Model.In this method,the spectral signal is sparsely reconstructed,the spectral matrix is used as the sensing matrix,and the predicted variable is the observed variable.During each iteration,the inner product value in the current sensing matrix and the observed matrix is calculated to find the largest The atomic vector of,remove it from the observation matrix,calculate a new residual,and ensure that the obtained residual is orthogonal to the set of atoms selected for each time,and finally reconstruct the original signal through multiple iterations The estimated value ensures the optimality of the iteration,reduces the number of iterations,realizes the dimensionality reduction of the spectral data and improves the prediction accuracy of the analysis model.(4)In order to verify the effectiveness of the OMP algorithm,combined with the characteristics of near-infrared spectroscopy,it was applied to the near-infrared spectroscopy analysis of the quality detection of Gentiana chinensis.WT with better prediction effect was selected as the spectral pretreatment method and combined with the OMP algorithm to establish a quantitative analysis model for the content of gentiopicroside and loganinic acid,and WT-SPA-MLR,WT-CARS-PLSR,The prediction results of the PLSR quantitative analysis model were compared.The results showed that the RMSEP of the quantitative analysis model established by WT-OMP for the contents of gentiopicroside and loganin in Gentiana were 0.1190,0.0936,and R~2were 0.9799,0.9845,respectively,compared with the prediction effects of the other three methods.With the improvement,the detection of the two main components in Gentiana chinensis obtained better prediction results.It can be concluded that the analysis model based on the WT-OMP algorithm can effectively improve the accuracy of the quantitative analysis of Gentiana chinensis and reduce the complexity,which is an effective quantitative analysis method.