Research On The Prediction And Analysis Of Peanuts Storage Quality Based On Electronic Nose

China is the world’s largest producer of peanuts.Peanut is rich in protein,fat,amino acids and vitamins.It has been loved by the broad masses.However,the fatty and protein contained in peanuts are easy to denature under the condition of elevated moisture and strong metabolism,especially to the high temperature and extreme humidity in summer,affected by the effect of post-ripening,release more moisture and heat,easy to fever,mold,oil and rackness,resulting in the deterioration of the quality of peanuts in storage,resulting in large losses,thus affecting the production,processing and export trade of peanuts.It is urgent to check the quality of peanuts from time to time.As an artificial olfactory device for analyzing,detecting complex odors and volatile components,the electronic nose has been widely used in the food industry for non-destructive testing,quality control,origin identification and shelf-life prediction.In this study,three objectives were emphasized:(1)sensory evaluation and physicochemical detection were used to analyze the quality changes of peanut in storage,and the e-nose detection was used to obtain the odor information of peanut kernel.(2)based on peanut quality classification label,support vector machine(SVM),k-nearest neighbor learning(KNN),decision tree,linear discriminant analysis(LDA)were used to establish the prediction model of peanut kernel.(3)using different conditions of peanut kernel to verify applicability of the storage quality prediction model.The main conclusions of the research work are as follows:(1)Sensory attribute evaluation was used to determine the sensory quality of baisha aged peanuts,baisha fresh peanuts and high oleic peanuts at 25℃.Physicochemical tests were used to determine the acid,hydrogen peroxide,moisture,and fatty acid content of the peanuts.Peanuts in different storage period are classified into three types.For baisha aged peanuts,test 1 was classified as type Ⅰ,tests 2 and 3as type Ⅱ,and tests 4 to 10 as type Ⅲ.For baisha fresh peanuts,test 1 was classified as type Ⅰ,test 2 to 4 as type Ⅱ,test 5 to 10 as type Ⅲ.For high oleic peanuts,test 1was classified as type Ⅰ,test 2 to 5 as type Ⅱ,and test 6 to 10 as type Ⅲ.(2)E-nose detection was used to determine various kinds of peanuts under storage periods.In order to achieve excellent prediction,the collected e-nose signals were optimized.In addition,the storage quality classification labels of peanuts are determined by sensory and physicochemical labels obtained from sensory and physicochemical detection.To establish the storage quality prediction model of peanut kernel,we contrast unsupervised and supervised algorithms.For baisha aged peanuts samples,the optimal prediction rate was 100% by using decision tree and k-nearest neighbor learning(KNN)classification model.For baisha fresh peanuts,the optimal prediction rate was 100% by using decision tree,linear discriminant analysis(LDA),and k-nearest neighbor learning(KNN)classification model.For high oleic peanuts,the optimal prediction rate was 100% by using k-nearest neighbor learning(KNN)classification model.(3)To study the applicability of the established model for prediction of peanut kernel,choose baisha fresh peanuts and high olic peanuts at 35℃.Sensory analysis,physicochemical detection,and e-nose detection were also performed to obtain a set of classification labels.The storage quality of baisha fresh peanuts and high oleic peanuts at 35℃ was divided into three types.Then we verify the predictive effect of the established mass model and analyze the applicability of the storage quality prediction model.For baisha fresh peanuts stored at 35℃,k-nearest neighbor learning(KNN)model,which has the best classification accuracy among the high oleic peanuts storage quality prediction models,was used to forecast the peanuts at normal temperature,and the prediction accuracy was 84%.