Modeling the Physiological Parameters of Brewer’s Yeast during Storage with Natural Zeolite-Containing Tuffs Using Artificial Neural Networks

Abstract

Various methods are used to prevent the deterioration of the biotechnological properties of brewer’s yeast during storage. This paper studied the use of artificial neural networks for the mathematical modeling of correcting the biosynthetic activity of brewer’s seed yeast of the C34 race during storage with natural minerals. The input parameters for the artificial neural networks were the suspending medium (water, beer wort, or young beer); the type of the zeolite-containing tuff from Siberian deposits; the tuff content (0.5–4% of the total volume of the suspension); and the duration of storage (3 days). The output parameters were the number of yeast cells with glycogen, budding cells, and dead cells. In the yeast stored with tuffs, the number of budding cells increased by 1.2–2.5 times, and the number of cells with glycogen increased by 9–190% compared to the control sample (without tuff). The presence of kholinskiy zeolite and shivyrtuin tuffs resulted in a significant effect. The artificial neural networks were required for solving the regression tasks and predicting the output parameters based on the input parameters. Four networks were created: ANN1 (mean relative error = 4.869%) modeled the values of all the output parameters; ANN2 (MRE = 1.8381%) modeled the number of cells with glycogen; ANN3 (MRE = 6.2905%) modeled the number of budding cells; and ANN4 (MRE = 4.2191%) modeled the number of dead cells. The optimal parameters for yeast storage were then determined. As a result, the possibility of using ANNs for mathematical modeling of undesired deviations in the physiological parameters of brewer’s seed yeast during storage with natural minerals was proven.