Identification Of The Origin And Content Analysis Of Chines Herbal Medicine Panax Notoginseng Based On Terahertz Time-domain Spectroscopy

Panax notoginseng is a traditional and valuable Chinese herb.Due to its high medicinal value and high demand,the Chinese medicine market suffers from the problem of substandard quality and insufficient content of related medicines.The current traditional testing methods for determining the quality of Panax notoginseng have lagged behind the development of the Chinese herbal testing market,and it has become a topic of interest for researchers to achieve rapid and effective detection of Panax notoginseng.Terahertz Time-Domain Spectroscopy(THz-TDS)is a new spectroscopic characterisation technique that has become a new tool for the rapid analysis of Chinese herbal medicine with its unique characteristics of high penetration,fingerprint spectral properties and high safety.In this thesis,it is based on the THz-TDS technology,and on the basis of the spectral characteristics of Panax notoginseng,combined with machine learning methods to established Panax notoginseng origin identification and content detection model.This provided a fast,efficient and accurate method for the quality detection of Panax notoginseng.The details of the study are as follows:(1)Study on the identification of the origin of Panax notoginseng.Panax notoginseng has regional characteristics and the price of Panax notoginseng varies greatly between different origins,with Wenshan in Yunnan having the highest value,and there is the phenomenon of Panax notoginseng from other origins masquerading as Wenshan Panax notoginseng.To address the issue of similar THz spectral characteristics of Panax notoginseng from different origins.This thesis provided a high performance Panax notoginseng origin identification model by improving the parameter optimization method based on Support vector machine(SVM).To address the problems of low global search efficiency in the early stage of the parameter optimization algorithm and slow convergence speed,the nonlinear coefficients a and Lévy flight are introduced to propose an improved Whale Optimization Algorithm(WOA)with high convergence and high global optimisation seeking efficiency,thus established the LAWOA-SVM model.The experimental results show that the model could effectively and quickly identify four different origins of Panax notoginseng with a fast convergence rate,and the correct identification ability could reach over 98%.(2)Study on the content analysis of Panax notoginseng.Panax notoginseng is expensive and in high demand in the market,there is demand exceeds supply and insufficient Panax notoginseng content in Prepared Chinese Medicine.To address the problem that the complex composition of Chinese herbal compounds leads to serious overlap of THz spectra and the samples do not carry obvious characteristic information.In this thesis,two-dimensional correlation spectroscopy(2DCOS)was introduced to obtain terahertz 2DCOS simultaneous spectra by applying external thermal perturbation to induce dynamic changes in the spectral signals.The experimental results show that compared to traditional THz absorbance spectroscopy data,the terahertz 2DCOS simultaneous spectrograms increase the intensity of weak peaks,making the different ratios of trichome information more distinctive features.Meanwhile,Competitive adaptive reweighted sampling(CARS)was used to provide a fast and accurate feature extraction means for Support vector regression(SVR),thus established a 2DCOS-CARS-SVR quantitative model.The model effectively improved the prediction accuracy of the Panax notoginseng content model,with the correlation coefficient and root mean square error of the content prediction set of Panax notoginseng samples being 0.9915 and 0.8160%,respectively.