CONSUMER PREFERENCE RESEARCH IN FORMING RATIONAL COMPONENT OF REGIONAL BRANDS IN THE MEAT PRODUCT MARKET

Abstract

The use of α-lactoglobulin hydrolysate produced with the use of Flavorpro 750MDP and Promod 439L enzyme preparations to obtain fermented milk drinks having reduced residual antigenicity has been suggested. It is necessary to provide normalized physical-chemical and acceptable organoleptic characteristics of resulting fermented milk drink. The purpose of the research is to study the possibility of applying the method of artificial neural networks for prediction of properties, quality indices and safety factors of normalized dairy mixes used to obtain low-allergenic fermented milk drinks. Organoleptic characteristics, physical and chemical properties of the normalized mixes have been studied with sensorymetric method. The method of evaluation of quality indices of normalized mixes and fermented milk drinks has been adapted using the method of artificial neural networks. Three-layer neural network with 6 neurons in the input layer, 12 neurons in the inner layer and 4 neurons in the output layer according to the number of output parameters has been used. The algorithm of back-propagation errors has been applied for training the network. The research results confirm that the obtained neural network predicts the main characteristics of normalized mixes with α-lactoglobulin hydrolysate almost accurately; the relative error does not exceed 2.6% when predicting α-lactoglobulin content, 3.9% when predicting residual antigenicity and 3.1% when predicting titratable acidity and organoleptic characteristics. This method is applicable for assessing the quality of finished goods and can replace the routine methods of analysis in force at the enterprises of dairy industry.

Keywords

• artificial neural networks
• quality index evaluation
• low-allergy fermented milk drinks