Study On Discrimination Method Of Salvia Miltiorrhiza By Artificial Neural Network Combined With Spectral Analysis

The identification and supervision of traditional Chinese medicine(TCM)is a key factor to ensure the clinical efficacy,human’s healthy and the steady development of the Chinese medicine industry.Hyperspectrum and laser-induced breakdown spectroscopy(LIBS),as two spectral analysis technologies,have the advantages of rapid real-time analysis,non-destructive or micro-damage detection,and can provide effective technical support for the rapid evaluation of the quality of TCM.However,the effective extraction and accurate analysis of trace multi-component spectral characteristic information in complex TCM systems and in the case of massive spectral data sets are still one of the current difficulties.This paper takes the identification,analysis and quality evaluation of TCM-Salvia miltiorrhiza as the practical need,and taking three aspects of Salvia miltiorrhiza from different sources,the geographical origin of Salvia miltiorrhiza,and the manufacturer of compound Salvia miltiorrhiza tablets as the research points,the quality identification method of Salvia miltiorrhiza combined with artificial neural network(ANN)and spectral analysis was studied,hoping to lay a theoretical foundation and technical support for accurate quality analysis and intelligent evaluation of Salvia miltiorrhiza.The main research work of the full paper is as follows:Firstly,Hyperspectrum method combined with artificial neural network(ANN)method was established to identify Salvia miltiorrhiza samples from different origins.In the experiment,the hyperspectra of 9 kinds of Salvia miltiorrhiza samples from different origins were collected.A supervised classification model was established by ANN method combined with 5 different spectral preprocessing methods.The mean centering(MC)method was selected as the best preprocessing method by comparison,and the classification model was built based on the ANN method;Compared with the k-nearest neighbor(KNN)method,the MC-ANN model has a better effect on the discrimination of Salvia miltiorrhiza from different origins,with a test set classification accuracy of 98.77%,and has higher sensitivity,precision and specificity.The overall results showed that hyperspectral combined with artificial neural network was a promising method for the analysis and identification of Salvia miltiorrhiza.Secondly,the method for discriminating Salvia miltiorrhiza samples from different geographical regions was developed by using laser-induced breakdown spectroscopy(LIBS)coupled with convolutional neural network(CNN).The LIBS spectra of Salvia miltiorrhiza samples from six geographical origins were collected and preprocessed with the maximum minimum normalization method.The classification model for discriminating these samples was then developed by using one-dimensional convolutional neural network.The discrimination accuracy of the developed CNN model reached 97.09%.Compared with support vector machine and k-nearest neighbour methods,the CNN model showed higher discrimination accuracy.The result demonstrates that the combination of LIBS and CNN is a practicable method for discriminating Salvia miltiorrhiza from different geographical regions.Finally,taking compound Salvia miltiorrhiza tablets as the research object,the diacrimination method of different manufacturers of compound Salvia miltiorrhiza tablets were established by hyperspectral and ANN method.The hyperspectral data of compound Salvia miltiorrhiza tablets from 7 manufacturers were collected;Through the comparison of five different spectral preprocessing methods,the multiple scattering correction(MSC)method is selected as the proper preprocessing method,and the discrimination model is built based on ANN;PCA method has also been applied to the established ANN model;The discrimination accuracy of the developed PCA-ANN model reached 98.41%,and has a higher sensitivity,precision and specificity.Therefore,the method of hyperspectral combined with ANN is accurate and reliable,and provides a new method for the quality standard supervision of compound Salvia miltiorrhiza tablets.