Neural observer to trehalose estimation

These It is generally accepted in yeast production industry that intracellular trehalose is an indicator of yeast fermentation capacity and viability. The disaccharide trehalose is a cytoplasmic compound, so it must be quantified after extraction by means of an off-line analytical method during a biomass production process. Thus, knowing experimental determinations of yeast trehalose content is always delayed; hence no opportune actions can be implemented in order to lead the production process toward a high intracellular trehalose concentration in the produced biomass. An attempt of predicting trehalose concentration in yeast cells through two different mathematical approaches is presented. On the one hand, a biomass and trehalose concentrations estimator was developed with a differential neural network technique. On the other hand, a structured model results are analyzed for explaining the main metabolic events that induce a trehalose accumulation in cells. Our results allow us to think that the coupling of both methods can provide acceptable information aimed at reaching high trehalose content in yeast. Indeed, by integrating the two alternatives, a trehalose-enriched yeast production process could be successfully driven.