COMPARISON OF STATE ESTIMATION TECHNIQUES FOR BIOTECHNOLOGICAL PROCESSES

In this article, application of state estimation techniques for indirect measurement of biomass concentration and specific growth rate in a typical fed-batch biotechnological process is discussed. Two state-of-the-art algorithms were implemented: Extended Kalman Filter (EKF) and feed-forward artificial neural networks (ANN) of various structures. Oxygen uptake rate (OUR), carbon dioxide production rate (CPR), and base consumption rate (BCR) along with their integrated quantities were used as direct reference measurements. The influence of measurement noise, experimental database size used for the model identification, and of the utilized input variables on the estimation quality was investigated. Recommendations for the application of the elaborated soft-sensors are given.