Observer design for nonlinear systems described by multiple models

This paper addresses both analysis and observer design for nonlinear systems modelled by decoupled multiple models. With respect to classic used multiple models, the decoupled multiple model is characterized by heterogeneous submodels in the sense that their state spaces may be of various dimensions. Thanks to this fact, flexibility and generality is introduced in the modelling stage. The main contribution of the paper is the development of new sufficient conditions on LMI form for ensuring the exponential convergence towards zero of the estimation error in the continuous and in the discrete-time. The new proposed conditions enable to obtain a better decay rate with respect to the existing conditions. The validity of the proposed methodology and its application to sensor faults detection and isolation is illustrated by an academic example.