Research On Near Infrared Technology For Rapid Detection Of Key Parameters Of Wine Brewing Process

With the rapid growth of Chinese economy,the wine industry began to develop rapidly.However,Chinese wine market competition environment confusion and quality problems emerging one after another Seriously hindered the rapid development of the wine industry.Therefore,it is urgent to develop a fast,efficient and accurate analytical technique to improve the quality control of wine making enterprises.In this paper,Near Infrared Spectroscopy(FTIR)and Chemomertrics were used to study the rapid and quantitative determination of key components in grape wine,grape wine fermentation broth and wine production,it is to provide theoretical guidance for improving the quality of wine making process.The main research contents are as follows:(1)Study on the rapid quantitative determination of total sugar,total acid,tartaric acid and malic acid in grape wine.The first derivative(FD),two derivative(SD),standard normal transformation(SNV)and multiple scattering correction(MSC)of four kinds of pretreatment methods are used to select the optimum extraction of principal components and process data,establishing partial least squares regression model.The result shows that the total sugar index by using the standard normal transformation(SNV)pretreatment method,the model is best when the number of principal components is 7,its standard deviation coefficient R2,verification RMSEP,relative error analysis of(RPD)were respectively 0.919、2.528、3.21 decided to test.The total acid,tartaric acid and malic acid index were treated by multiple scattering correction(MSC),and the model is best when principal components were 6,7 and 5 respectively,the determination coefficient R2 was 0.921,0.902,and 0.906,the standard deviation of RMSEP was 0.486、0.475 and 0.305,the relative error(RPD)was up to 3.05、3.04 and 3.08,respectively.The results show that the near infrared spectrum analysis technology is feasible and effective for the determination of the main components of wine grape.(2)A rapid and quantitative method for the determination of total sugar and alcohol content in wine fermentation broth was studied.The combination of interval least squares(SiPLS),moving window partial least squares(MWPLS)combined with genetic algorithm(GA)was used to filter wavelength segment and reduce the number of variables.The model is optimized by four pretreatment methods:the first derivative(FD),the second derivative(SD),the standard normal transformation(SNV)and the multiple scattering correction(MSC).On the basis of the standard normal transformation(SNV)method,the partial least square model of SiPLS-GA screening is more ideal,R2 reached 0.944 and 0.957,RMSEP was 0.266 and 0.162,RPD was 4.36 and 5.12,and the best principal component was 6 and 5,respectively.The results show that the variable selection method can reduce the modeling complexity and improve the accuracy and stability of the model.(3)A rapid quantitative method for the determination of main indexes of wine was studied,the characteristic wavelength of CARS is used as the input variable of radial basis function(RBF)neural network.Near Infrared quantitative model has established Wine in total sugar,total acid,dry extract,volatile acid,and compared with the partial least squares regression model results,for determining the best modeling method.The results show that the radial basis function(RBF)neural network model is better than the PLS model for the total sugar,total acid,dry extract,volatile acid quantitative model,R2、RMSEP and RPD value is superior,R2 reached 0.925,0.945,0.928,RMSEP reached 0.305,0.223,0.236,0.063 and RPD reached at 3.81,3.99,5.21,5.08,respectively.The results show that the model can be applied for rapid detection of total sugar,total acid,Wine dry extract,volatile acid content,which provide a theoretical basis for the real-time and accurate monitoring of the changes of the main indexes of wine and guide the production process of wine scientifically.