Iterative Identification for Control and Robust Performance of Bioreactor

A main result of control oriented identification is that, to minimize the closed-loop error that will ultimately result from a model-based controller, the identification of the model should use closed-loop data produced with the very same controller. However, the controller is not known prior to the actual experiment and therefore iterations between controller design and closed-loop identification are performed. This paper presents a modified procedure based upon the works of de Callafon (1998) and Tay et al. (1998). The performance of the above procedure is investigated on a continuous bioreactor which may exhibit unstable operation just beyond the critical dilution rate.