Font Size: a A A

Multi-way Pattern Recognition Is Used In The Study Of Origin Traceability Of Traditional Chinese Medicine

Posted on:2020-05-04Degree:MasterType:Thesis
Country:ChinaCandidate:S MaFull Text:PDF
GTID:2381330578951355Subject:Chemistry
Abstract/Summary:PDF Full Text Request
The quality of Traditional Chinese Medicine is affected by many factors,one of which is more important is its origin,so it is necessary to identify the origin of Chinese medicine.Firstly,this paper introduces the significance,methods and research status of Traditional Chinese Medicine production in the introduction part,it is found that most of the current research on the origin identification of Traditional Chinese Medicine is based on two-dimensional data level,and the three-dimensional data contains more sample information to provide more reference information for the identification.Therefore,this paper will combine three-dimensional data with chemometrics multi-dimensional pattern recognition method to identify the origin of Traditional Chinese Medicine research.In this paper,three Traditional Chinese Medicines,kudzu root,bletilla striata and panax quinquefolium L were selected as research objects,and several multi-dimensional pattern recognition methods were used to classify and identify their origins.The specific contents are as follows:In the second chapter,the three-dimensional fluorescence spectra of pure kudzu root from different geographical regions and five adulterated kudzu root with different adulteration ratios were measured.The data were analyzed and processed by two multi-dimensional pattern recognition methods: multi-dimensional principal component analysis(M-PCA)and Ndimensional partial least squares discriminant analysis(N-PLS-DA).The results showed that M-PCA and N-PLS-DA had high accuracy in classification and recognition of kudzu root from from different geographical regions.When classifying adulterated kudzu root,the result of NPLS-DA is better than that of M-PCA.In the third chapter,the three-dimensional fluorescence spectra of bletilla striata from different regions were determined.Firstly,the data set was analyzed by self-weighted alternating trilinear decomposition(SWATLD)to obtain the spectra and relative concentrations of the four components in white and medium.Those Information can be used for identify the origin and quality control.In addition,the four classification methods: M-PCA,N-PLS-DA,unfolded partial least squares discriminant analysis(U-PLS-DA),self-weighted alternating trilinear decomposition combined with partial least squares discriminant analysis(SWATLDPLS-DA)are used to identify the origin.The results show that the total accurate recognition rate of N-PLS-DA is 98.9%,and the total accurate recognition rate of U-PLS-DA and SWATLD-PLS-DA is 100%.In the fourth chapter,the three-dimensional fluorescence data of panax quinquefolium L from different origins were measured,and then the three-dimensional fluorescence data of panax quinquefolium L were analyzed and processed by M-PCA and N-PLS-DA.The results showed that the M-PCA showed that the panax quinquefolium L from different origin had their own clustering trend,but the overall classification result was not good.The N-PLS-DA has a good classification recognition effect.In the training set,the accurate identification rate of panax quinquefolium L from each place of origin has reached 100%.In the prediction set,the accurate identification rate of panax quinquefolium L from each place of origin has also reached 98%,and the overall accurate identification rate has reached 99%.The above content indicates that three-dimensional fluorescence spectrum combined with multi-dimensional pattern recognition method is a feasible method to identify the origin and adulteration recognition of traditional Chinese medicine,and it is also simple,fast and green.
Keywords/Search Tags:Traditional Chinese Medicine, Origin identification, Multi-dimensional principal component analysis, N-dimensional partial least squares discriminant analysis, Self-weighted alternating trilinear decomposition
PDF Full Text Request
Related items