Recursive Output Error Identification Algorithm for Switched Linear systems with Bounded Noise

Abstract : The problem considered in this paper is the identification from input/output data of Switched linear systems with Switched Output Error (SOE) models. The identification is carried out in the case where an upper bound on the noise is assumed to be known. This problem includes the estimation of the discrete state that corresponds to the active sub-model and the estimation of the parameters of all sub-models. The proposed algorithm proceeds in two successive stages. The stability and convergence proprieties are established.