Traceability Of Soybean Origin In Different Provinces Of China Based On Mineral Element Fingerprint Analysis Technology

As an important basic food crop in China,soybean is widely cultivated in China.China has excellent soybean quality and is favored by consumers.However,due to problems such as adulteration and sub-charging,the interests of consumers and their health are impaired.Therefore,it is of great significance to study the origin tracing technology of soybean origin in different provinces of China.This study is mainly to find the traceability index of mineral elements in soybean production areas in different provinces of China,in order to improve the accuracy and stability of the application of mineral element fingerprint analysis technology in traceability of soybean production areas.And discussed the feasibility of the application of portable energy dispersive X-ray fluorescence spectroscopy(ED-XRF)technology in mineral element fingerprint analysis technology.In this study,inductively coupled plasma mass spectrometry(ICP-MS)was used to analyze a variety of mineral elements in 296 soybean samples from major soybean producing areas in 5 provinces of China(Heilongjiang,Liaoning,Inner Mongolia,Xinjiang,Henan).The results of analysis of variance show that there are significant differences among the five major producing areas in the nine mineral elements Mg,P,K,Ca,Mn,Fe,Cu,Zn and Rb(p <0.05);The results of principal component analysis show that the mineral elements obtained based on ICP-MS method are analyzed by principal components to obtain three principal components,and the cumulative variance contribution rate is 72.960%.The principal component 1 is mainly composed of Mg,P,K,Mn,Fe,Zn,the principal component 2 is mainly composed of Ca and Rb,and the principal component 3 is mainly composed of Cu;However,principal component analysis can not distinguish the five origins very well.In this study,a multilayer perceptron was used to establish a soybean origin classification model.By optimizing the number of neurons in the hidden layer,the structure of 9-5-5 was finally determined For the origin classification model,the accuracy of the training set is 100%,the accuracy of the test set is 98.3%,and the holdout set is 98.4%.It proves that the origin classification model established by these 9 mineral elements and the multi-layer perceptron can successfully classify soybeans from different provinces of China.In this study,the irradiation time of the portable ED-XRF instrument was optimized,and it was found that the instrument's detection efficiency(detection speed,instrument precision)was the best at 300 s sample irradiation time.The standard curves of 9 elements Mg,K,Ca,Mn,Fe,Ni,Cu,Zn,Rb were established by the standard sample method.The optimized instrument analyzes a variety of mineral elements in the above samples,and the results of analysis of variance show that there are significant differences among the five major producing areas: the 9 mineral elements Mg,K,Ca,Mn,Fe,Ni,Cu,Zn,Rb;The results of principal component analysis show that the mineral elements obtained based on the portable ED-XRF technology are analyzed by principal components to obtain three principal components,and the cumulative variance contribution rate is 65.943%.The principal components 1 is mainly composed of Cu,Zn and Ca,the principal components 2 is mainly composed of Mg,and the principal components 3 is mainly composed of Mn and Ni;A multi-layer perceptron was used to build a soybean origin classification model,optimize the number of hidden layer neurons,and determine a 9-5-5 origin classification model.The accuracy of the model training set was 99.4%,the test set was 94.1% and the holdout set was 96.5%.This proves that the application of portable ED-XRF technology in the traceability research of food origin is feasible.