A comparison of three sets of DSP algorithms for monitoring the production of ethanol in a fed-batch baker’s yeast fermenter

This paper aims at presenting the design approach of three sets of DSP algorithms for a fed-batch biochemical reactor. The proposed algorithms are based on the estimation/observation methods enhanced with artificial neural networks. Here, three popular estimation schemes, namely extended Luenberger observer, extended Kalman filter and adaptive state estimator, have been designed to estimate the specific growth rate, substrate consumption rate and product formation rate on the basis of measured process state variables. The neural network model infers the biomass concentration. It is supposed that the substrate and product concentrations along with the broth volume are measured quantities. The comparative performance of the proposed DSP schemes has been inspected through simulation results dealing with a fed-batch baker’s yeast fermenter.