Research On Liquor Blending Control Method Based On Bionic Taste Evaluation

The nature of liquor is extremely special.Except for individual manufacturers that use some automatic detection devices in the liquor blending process,other manufacturers still use traditional manual tasting methods to determine the optimal liquor blending formula.Manual tasting methods are subjective and will affect the determination of blending formulas.Therefore,improving the level of automation in the blending process of liquor has become an urgent problem to be solved.Electronic tongue technology is a bionic taste evaluation method developed in recent years.It provides a brand new solution for the determination of the optimal formula of liquor blending,and improving the automation level of liquor blending production.In this paper,electronic tongue technology is used to realize the classification,identification and differentiation of basic liquors.Establish an association model between the electronic tongue response characteristics and the physical and chemical components of liquor.On this basis,an optimal control method for liquor blending is proposed based on genetic algorithm and particle swarm algorithm.main tasks as follows:(1)Realize the distinguishing and identification of different kinds of liquor.The ability of electronic tongue technology to distinguish different types of liquor is a prerequisite for liquor blending.Use the potential electronic tongue to collect the solution information of different liquors,and distinguish and identify different liquors through PCA and DFA methods.Use voltammetric electronic tongue,choose cyclic voltammetry and differential pulse voltammetry as excitation potential,use principal component feature extraction and wavelet packet transform to preprocess the collected solution information,and combine the LDA algorithm to compare the classification results under different conditions.Under the optimal conditions,the CA algorithm is used to further realize the distinguishing and recognition of different kinds of liquor.(2)Construct a quantitative prediction model of liquor flavor.The linear relationship between the corresponding characteristics of electronic tongue and the physicochemical composition of liquor is the basis for liquor blending.Using GC-MS,the relative content of the physical and chemical components that affect the flavor of Luzhou-flavor liquor is studied from four aspects:esters,alcohols,acids and hydroxyl compounds.Taking the response characteristics of the volt ampere electronic tongue and the potential electronic tongue as independent variables.The relative content of physical and chemical components is used as the dependent variable.Use ELM and WNN two neural network algorithms to build predictive models.Deeply explore the inner relationship between the response characteristics of the electronic tongue and the relative content of the physical and chemical components of the liquor.(3)Design the best liquor blending formula.Taking 4 kinds of Luzhou-flavor liquor as the base liquor and 1 kind of Luzhou-flavor liquor as the target liquor.according to different combinations of base liquor,cost and dosage are taken as objective functions.electronic tongue response characteristics are used as constraints.Combining genetic algorithm and particle swarm algorithm to control the blending ratio while finding the optimal blending formula under the conditions of the lowest cost and the least amount.Use GC-MS technology to verify whether the physical and chemical composition indexes of the optimal blending formula are close to the physical and chemical composition indexes of the target wine.This paper proposes an automatic control method for the optimal formula of liquor blending.It realizes the rapid and accurate screening of the blending formula during the liquor blending process.It provides a new solution for the guarantee of flavor quality and the improvement of automation level in the production process of liquor.