An algorithm for real-time estimation of Mezcal fermentation parameters based on redox potential measurements

We describe an algorithm for the continuous monitoring of the biomass and ethanol concentrations as well as the growth rate in the Mezcal fermentation process. The algorithm performs its task having available only the on-line measurements of the redox potential. The procedure combines an artificial neural network (ANN) that relates the redox potential to the ethanol and biomass concentrations with a nonlinearobserver-based algorithm that uses the ANN biomass estimations to infer the growth rate of this fermentation process. The results show that the redox potential is a valuable indicator of the metabolic activity of the microorganisms during the Mezcal fermentation. In addition, the estimated growth rate can be considered as a direct evidence of the presence of mixed culture growth in the process. Usually, mixtures of microorganisms could be intuitively clear in this kind of processes, however the total biomass data do not provide definite evidence by themselves. In this paper, the detailed design of the software sensor is presented, as well as its experimental application at the laboratory level.