Study On The Origin Traceability Of Mung Beans Based On Raman And NIR Spectroscopy And The Influencing Factors

The origin traceability technology of agricultural products can provide technical and theoretical basis for the traceability and confirmation of geographical indication products and regional famous and special products.At present,Raman and Near-infrared spectroscopy have been widely studied for the origin traceability discrimination of agricultural products,and their two techniques are different in terms of traceability effects.To further investigate the discrimination of mung bean origin based on Raman and Near-infrared spectroscopy,this experiment was conducted with 120 geographical indication mung beans from Tailai and Dulbert Mongol Autonomous County of Heilongjiang Province,Baicheng City of Jilin Province,and Sishui County of Shandong Province in 2021,and the mung bean samples were collected by Raman and Near-infrared spectra through micro confocal Raman spectrometer and Fourier diffuse reflectance Near-infrared spectrometer,and after the collection,different spectral pre-processing methods were applied to the original spectra of mung beans,and principal component analysis was used to reduce the dimensionality of the spectral data,and the effect of different mung bean origin traceability based on Raman and Near-infrared spectroscopy was compared and analyzed by combining chemometric methods,and the influence of sample quantity,sample traceability range and sample shelf life on the origin traceability results of different mung beans was investigated.It provides new ideas for the research of origin traceability of agricultural products.The specific research results are as follows:（1）Study on the origin traceability of mung beans of different origins by Raman spectroscopyIn order to study the results of traceability discrimination of different mung bean origins based on Raman spectroscopy.The Raman spectra of mung bean samples of different origins were analyzed by principal component analysis,and after a series of spectral pre-processing,the total cumulative variance contribution of the first three principal components reached 92.30%>80.00%,and the Raman discrimination model based on mung bean samples of different origins was established by combining the K-nearest neighbor method,and its overall correct discrimination rate was 76.00%.Meanwhile,the discriminative model constructed based on Raman spectral data of mung beans of different origins was analyzed and evaluated,and the accuracy of its model was 75.75%,and the F₁-score value was above 0.66,which indicated that the effect of Raman spectroscopy technology for the discrimination of different mung bean origins was low,and it was difficult to achieve the effect of practical application in the study of traceability of different mung bean origins.（2）Study on the origin traceability of mung beans of different origins by Near-infrared spectroscopyTo investigate the results of the traceability discrimination of different mung bean origins based on Near-infrared spectroscopy.A comparison of six pre-processing methods using data standardization（Standardization）,normalization,mean centralization（CT）,and standard normal variance transformation（SVM）was performed by using PCA analysis.Among them,the data normalization method has the largest total cumulative contribution among all NIR spectral pre-processing methods,and the total cumulative variance contribution of the first three principal components reaches 98.16%,respectively,which is 5.86%higher than the total cumulative variance contribution of the three principal components after Raman spectral pre-processing.Combined with the K-nearest neighbor method to establish the NIR discrimination model based on mung bean samples of different origins,the overall correct discrimination rate for mung bean samples of different origins was 98.67%.At the same time,the discriminant models based on the NIR spectral data of mung beans from different origins were analyzed and evaluated,and the accuracy of the models was 98.25%,and the F₁-score values were above 0.97,which was 0.31%higher than the F₁-score values of the discriminant models based on Raman spectral data.Therefore,compared with Raman,Near-infrared spectroscopy is more suitable for origin discriminative studies on mung bean samples of different origin sources,but there are still influenced by many factors,such as the number of samples,sample traceability range and sample shelf life.（3）Study on the effect of sample size on the correct discrimination rate of mung bean NIR origin tracingTo further analyze the effect of sample size on the origin traceability discrimination results.The NIR discrimination model was established by selecting small（n=200）,medium（n=400）,and large（n=600）samples.The results showed that the overall origin discrimination rate of mung beans using the number of samples（n=600）was 98.67%,which was 4.67%higher than the overall origin discrimination rate of mung beans with a small number of samples（n=200）;meanwhile,the accuracy of the origin traceability discrimination model established using the number of samples（n=600）was 98.25%,which was higher than that of the origin discrimination model established with a small number of samples（n=200）.The accuracy of the origin traceability model established with a small number of samples（n=200）increased by 5.25%.This indicates that the higher the number of samples collected,the higher the accuracy of the discrimination of mung bean samples of different origins.（4）Study on the effect of sample traceability range on the discrimination rate of mung bean NIR origin traceabilityTo further analyze the influence of the traceability range of samples on the origin traceability discrimination results.By taking the origin of mung beans from Tai Lai as the reference point,the overall origin discrimination correct rate of mung beans with a larger traceability range（Tai Lai-Sishui）was selected to be 100.00%,and the overall origin discrimination correct rate of mung beans with a smaller traceability range（Tai Lai-Baicheng and Tai Lai-Dulbert Mongol Autonomous County）was improved by more than 2.00%;meanwhile,by comparing the evaluation indexes of origin discrimination models using the same conditions,the overall origin discrimination correct rate of mung beans with a larger traceability range（Tai Lai-Sishui）and a smaller traceability range（Tai Lai-Baicheng and Tai Lai-Dulbert Mongol Autonomous County）was improved by more than 2.00%.At the same time,by comparing the evaluation indexes of origin discrimination models using the same conditions,the accuracy of origin discrimination models established using mung bean samples with a larger traceability range（Tailai-Sishui）and a smaller traceability range（Tailai-Baicheng and Tailai-Dulbert Mongol Autonomous County）was improved.It indicates that it is difficult to accurately discriminate the origin traceability model based on a small range.（5）Study on the effect of shelf life on the discrimination rate of mung bean NIR origin traceabilityTo further analyze whether the shelf life of the samples could be used in the origin traceability model accuracy discrimination study.The NIR discrimination models were established by selecting mung bean samples with different shelf-lives（90 d,180 d,270 d,360 d,90-180-270-360 d）from five origins.The overall discrimination rate of the four origins based on different shelf life was more than 86.00%,and the model accuracy verification comparison of the origin discrimination of 20 randomly collected mung bean samples from outside the modeling of the four origins with different shelf life can be introduced that different sample shelf life is a factor affecting the establishment of the origin discrimination model.Therefore,the influence of as many shelf-lives as possible on the origin discrimination results of mung beans can be further investigated to provide a theoretical basis for the in-depth development of different high value-added agricultural products.