Comparison between Adaptive and Model-Based Extended Kalman Filters

The efficiencies of an adaptive Kalman Filter (AKF) and of a classical model-based Kalman Filter (EKF) are compared for the estimation of state variables in batch and fedbatch yeast cultures. Fully devoted to biomass estimation AKF behaves better that EKF in most cases except for the estimation of low residual glucose concentration in fedbatch, which can be a drawback in the establishment of control strategies based on estimated values of the state variables.