Robust state estimation for jump Markov linear systems with autonomous mode transitions

This paper addresses the robust state estimation problem for a class of jump Markov linear systems (JMLSs) with autonomous mode transitions. By describing the behavior of the autonomous mode transitions as Gaussian forms, we propose a novel robust state estimation algorithm by applying the basic interacting multiple model (IMM) approach and the H_∞ estimation technique. Moreover, as the performance of the H_∞ estimation depends on a group of weighting parameters, we present a way to tune them recursively. Simulation results show that the proposed algorithm tends to be more effective than the Kalman filtering counterpart when the noise statistics are not known exactly.