Analysis Of Characteristic Plants In Xinjiang By Near Infrared Spectroscopy

In this paper,near infrared spectroscopy analysis technology was used as the main means of detection,machine learning and deep-learning in chemometrics as the main data processing method,and Cymaiti apricot with the name of "Xinjiang First Apricot " as the research object of fruit non-destructive analysis.Soluble solid content(SSC)was used as the main research index.and Coreopsis tinctoria as the representative of medicinal plants for nutrient composition,source detection and wild/planting discrimination.A portable miniature device was used to model the SSC of Cymaititi apricot.Hundreds of calibration models were established by chemometrics means.Eighty-four 84 valid models were selected from them.By comparison,the optimal results were obtained.The results showed that reliable and stable quantitative prediction models can be established with the simplest instrument.The main research contents and conclusions are summarized as follows:1)In order to achieve the purpose of rapid and miniaturization detection of Cymaiti apricot soluble solids content(SSC),the SSC of Cymaiti apricot was measured by micro near infrared spectrometer and digital sugar meter,singular value was removed by Monte Carlo method.Then the modeling results including continuous wavelet transform(CWT),Savitzky-Golay(SG-smoothing),multiplicative scatter correction(MSC),standard normal variate(SNV),firstderivative(1st),second derivative(2nd)and the spectral pretreatment methods of two combinations were compared.Then the models of CWT,CWT+MSC and CWT+SNV showed good results.Based on above three pretreatment methods,we obtained the spectral variables of Cymaiti apricot SSC by comparing four variable selection methods including competitive adaptive reweighted sampling(CARS),randomization test(RT),informative variables elimination(MC-UVE)and C＿value.Through screening and comparison,we obtained the optimal partial least squares(PLS)prediction model for Cymaiti apricot SSC.The results showed that the rapid nondestructive analysis of Cymaiti apricot SSC could be accomplished by using the miniaturized equipment and proper spectral pretreatment and variable selection methods.2)In order to realize the rapid analysis of soluble polysaccharides in Coreopsis tinctoria,compare the rapid analysis effect of Fourier near-infrared spectrometer and micro-instrument and reduce the detection costs,we selected 10 kinds of Coreopsis tinctoria from different producing areas as research samples,and took soluble polysaccharides as the main research index.The soluble polysaccharides were extracted by wet method,and most of the impurities were removed by distillation and extraction methods,and glucose was used as the standard.The standard curve and equation of linear regression were established by UV visible spectrophotometer.Than the content of total polysaccharide in the extract of Coreopsis tinctoria was determined by concentrated sulfuric acid-phenol method.The spectral data of Coreopsis tinctoria were collected by Fourier table near infrared spectrometer and micro spectrometer respectively.According to the modeling process,fifteen spectral preprocessing methods were compared,and then the better prediction model was selected based on two NIR Spectrometer.The model established by Fourier transform spectrometer can basically accurately predict the polysaccharide content of Coreopsis tinctoria from 10 origins,while the micro device can only roughly predict the polysaccharide content of Coreopsis tinctoria from different origins based on the optimal variable.The model established by Fourier transform spectrometer can basically accurately predict the polysaccharide content of 10 producing areas,and the micro equipment can roughly predict the polysaccharide content of different producing areas.The results showed that near-infrared spectroscopy can be accurately used to predict the content of polysaccharides in Coreopsis tinctoria.The successful of the model indicated that the polysaccharides components in medicinal plants were expected to be rapidly and non-destructive analyzed by near-infrared spectroscopy.Because the research results were less than ideal,there is still much room for improvement in the future,which provides reference for the rapid analysis and determination of the quality of Coreopsis tinctoria from different producing areas.3)In order to realize the rapid identification of cultivated and wild Coreopsis tinctoria by near infrared spectroscopy,and reduce the phenomenon of "impostor" and shoddy in the market due to the low yield and high medicinal value of wild Coreopsis tinctoria,this paper adopted Principal component analysis(PCA)method was used to reduce the dimension of the near infrared spectral datas of the Coreopsis tinctoria,and the method of near infrared spectral pretreatment and variable selection methods were added,and a better classification result was obtained.The spectral data of unknown samples were put in,and the same methods were used for spectral processing,and a better classification result was obtained.The results showed that the method could identify both wild and different producing areas,and provide reference for further rapid detection of Coreopsis tinctoria market.