A SIMPLE IMPROVEMENT OF INTERVAL ASYMPTOTIC OBSERVERS FOR BIOTECHNOLOGICAL PROCESSES

A guaranteed interval estimation for the state can be the basis for a robust control strategy. In this paper we propose to combine two kinds of techniques to better estimate the interval for the state, especially in presence of uncertainties. Indeed, we use the hybrid observers as proposed by Lemesle and Gouzé (2005) that combine an asymptotic and a high gain observer. This observer assumes a loose bounding of the unknown function. We modify the observer to generate an interval observer. Each observer gain leads to bounds of the state. By taking the envelope of these various bounds we define a new estimate whose convergence is much better. We apply this new observer to a classical bioprocess model involving one biomass and one substrate and demonstrate that the convergence rate can be increased with respect to the interval asymptotic observer.