State estimation of nonlinear systems using multiple model approach

This paper addresses the problem of state estimation of nonlinear systems described by a Takagi-Sugeno multiple model with unmeasurable decision variables. The method is based on the reformulation of the multiple model in an equivalent form. First, the convergence conditions of the state estimation error are established using the Lyapunov method and they are expressed in LMI formulation. Secondly, performances of the observer are enhanced by pole clustering and L₂ attenuation of bounded exogenous disturbances. Finally, the method is applied to estimate the state of a link flexible joint robot.