| An object-oriented rule-based expert system application for continuous monitoring, analysis and control of fed-batch fermentation processes was designed, implemented and tested. With an efficient feedback control design and integration of the process equipment with the analysis equipment, it was possible to control substrate feed rate based on any of the following: mass-flow of the substrate, real-time analysis of concentration of substrate, respiratory variables from off-gas analysis, biomass from cell density probe or real time biomass estimation from a growth model. A fiber optic cell density probe that works on light scattering principle was used for cell density measurements in real time and also for feedback control of specific growth rate during the critical methanol feed phase of a Pichia pastoris fermentation process. A neural network based predictive model to estimate biomass from the real time measurable methanol metabolic variables was developed using the historical data consisting of wide variations in the process conditions. |
