Theoretical Study On The Detection Of Food Additives By Support Vector Machine And Fluorescence Spectrometry

With the rapid development of the food industry,a wide range of additives are applied to food,but some unscrupulous businesses over range,over use these,and even add the chemical reagents that are banned in our country in the food production process.Research and practical show that the use of non-standard additives will cause a great harm to the health of the human body,so how to accurately detect the types and content of food additives to ensure food safety can be a key issue.Through the analysis and research,the fluorescence method has the characteristics of high sensitivity,selectivity and simple operation.So in this paper,the concentration of potassium sorbate and sodium methyl p-hydroxybenzoate and the concentration of carmine in aqueous solution are studied by fluorescence method,and finally achieve the quantitative analysis of the material.The molecular structure of the three additives and the feasibility of fluorescence detection are analyzed by studying the mechanism of fluorescence emission.Fluorescence spectroscopy is used to study the fluorescence characteristics.The mixed solution of different concentration range are prepared in order to further study the relationship between the fluorescence intensity and the concentration of the mixture and providing the theoretical basis for the model establishment.In this paper,the least squares support vector machine model is applied to the regression analysis of fluorescence intensity and mixed matter concentration.In order to solve the problem of parameter selection,an improved particle swarm optimization algorithm and adaptive genetic algorithm are proposed to optimize the parameters.The experimental samples are trained to establish a regression model,and then to predict the unknown concentration of the samples.Based on the IPSO-LSSVM and AGA-LSSVM models,the experimental samples of the three additives are studied and analyzed.Comparing with BP neural network,PSOLSSVM and GA-LSSVM algorithm,the improved model performance is verified in terms of recovery rate and error.The results show that the two models have a unique advantage in the fluorescence detection of food additives,and the method of detecting the content of additives is further optimized.