State observers for reaction systems with improved convergence rate
Résumé
The problem of designing state observers for reaction systems whose convergence rate is faster than the standard asymptotic observers [7] is addressed in this paper. It is assumed that the reaction functions are known and that there are more measurements than “independent” reactions. If the unmeasurable state enters linearly in the reaction functions we propose an observer that converges in finite-time, under very weak excitation assumptions. If this dependence is nonlinear, we additionally assume that there is an element of the reaction functions vector that depends only on one unmeasurable state and that these functions are strictly monotonic. Under these conditions, a state observer that ensures exponential convergence of the states that appear in the reaction functions is designed. For the unmeasurable states that do not appear in these functions, the convergence is similar to the one of the asymptotic observers.