Design of robust Hinf observers for nonlinear systems using a multiple model
Résumé
A particular class of multiple model, known as decoupled multiple model, is used in order to cope with the state estimation problem of nonlinear systems. This attractive kind of multiple model is characterized by submodels of which the state belong to spaces of various dimensions, in contrast to the popular Takagi-Sugeno multiple model where the dimensions of the state space of the submodels are identical. Thus the decoupled multiple model is suitable for modelling complex systems with variable structure in the operating range and this fact offers promising prospects in the modelling, control and diagnosis of complex non linear systems. An original procedure for designing a proportional observer and a proportional-integral observer ensuring Hinf performances is proposed. Sufficient conditions for ensuring the estimation error convergence are derived employing the LMI framework. Comparison between the state estimation provided by both observers is given via a simulation example.
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...