| Boletus has delicious taste and rich nutritional value,which is deeply loved by consumers.As the consumption demand of wild bolete increase and the market price continues to grow,it has become one of the important economic sources for increasing farmers’income in mountainous areas.At the same time,food safety has attracted extensive attention.On the one hand,the geographical origin of agricultural products is a topic of concern,which has different effects on food quality and authenticity.Different soil background values in different regions lead to different enrichment degrees of heavy metals in Boletus.In addition,the complex terrain and diverse climate types in Yunnan Province provide an ideal environment for the growth and reproduction of wild fungi.However,wild fungi are widely distributed.Due to the influence of various ecological factors such as altitude,temperature,rainfall,and soil,the growth environment of bolete is very complicated.The diversity and complexity of chemical components make the quality of boletes vary significantly.On the other hand,in addition to the differences in the quality of different kinds of Boletus,Boletaceae contains some poisonous bolete,which brings certain dangers to people’s consumption.There are many kinds of wild edible mushrooms with high morphological similarity between species.It is difficult to accurately identify and classify them by traditional methods.Misidentification of toxic mushroom as edible and their consumption lead to poisoning events from time to time.Moreover,bolete is often dehydrated and dried into dry products to prolong the shelf life,and still maintain its flavor to a great extent.Dry food is not easy to rot,and it is difficult to distinguish the year from the appearance.With the extension of storage time,the nutritional components and flavor substances in bolete will gradually decrease,the content of amines will increase,and the quality of mushroom will decline.There are many phenomena in the market that the old and the new alternate.The illegal traders mixed the deteriorated dried products in the fresh bolete dried products,which is shoddy and infringes on the interests of consumers.In addition,With the development of science and technology,some efficient and intelligent modern technologies emerge in endlessly,and are widely used in various fields,which has brought convenience and breakthrough to life and scientific research.After successfully solving several challenges in the field of computer vision,deep learning is now expanding in the field of chemometrics.How to scientifically and quickly determine the origin,species and storage life of bolete is one of the main challenges facing the edible fungus market.In this study,1792boletes fruiting bodies were taken as research objects to identify the origin,species and storage life of boletes,and explore and verify the advantages of the deep learning model in bolete identification.(1)The content of heavy metals in wild edible fungi is closely related to the growth environment.There are significant differences in the Cd elements of Boletus bainiugan from different places of origin.The Cd content of B.bainiugan from Kunming was significantly higher than that of Chuxiong and Yuxi.The content of Cd in B.bainiugan was significantly correlated with the soil.In addition to the difference of heavy metal elements in bolete fruiting bodies in different regions caused by the soil background value,the adsorption capacity of different Boletus species to heavy metals is also different.In this study,the content of Cd in B.bainiugan was significantly higher than that of the other three Boletus.(2)This study was based on the determination of total phenolics content to evaluate the effects of species and environmental factors on the quality of boletes.Through multiple linear regression(MLR)analysis,bio16 and T＿GRAVEL were key environmental factors affecting the content of total phenolics.The R~2 of MLR equation was 0.604,which shows that the model was effective in predicting the total phenolics in bolete under different environmental conditions.(3)Qualitative and quantitative analysis of Lanmaoa asiatica in different storage periods based on Fourier transform near infrared(FT-NIR)and chemometrics.The results showed that with the prolongation of storage time,the chemical substances in the fruiting bodies were decomposed,resulting in an increase in the contents of uridine,adenosine and guanosine.The model was optimized by spectral preprocessing and feature variable extraction,which improves the predictability of the model.Partial least squares discriminant analysis(PLS-DA)could effectively identify samples with different storage years,and its model accuracy was 100%.In the established partial least squares regression(PLSR)prediction model for uridine,guanosine and adenosine,the regression coefficients R~2 were 0.9087,0.9010 and 0.7937 respectively,and the RPD values were 3.86,4.21 and3.12 respectively.In general,the method has good stability and applicability,and provides an effective and rapid analysis method for food safety and quality evaluation.(4)In this paper,different data pretreatment was carried out for the FT-NIR,and the modeling results of PLS-DA,support vector machines(SVM),random forest(RF)and residual neural network(ResNet)were compared.The results show that PLS-DA,SVM and RF models need a suitable combination of pretreatment for spectral data.The purpose is to improve the accuracy of the model and avoid over fitting.The ResNet model was established based on the original spectrum.The accuracy of the model was 100%,and there is no over fitting phenomenon.Compared with the above four models,ResNet is almost unaffected by data type,sample size and other factors,and has absolute advantage in Boletus species identification. |
