The extended state filter for a class of multi-input multi-output nonlinear uncertain hybrid systems

Abstract

This paper proposes a novel filter, the Extended State Filter(ESF), for a class of multi-input multi-output(MIMO) nonlinear systems with the continuous uncertain dynamics and the discrete measurement polluted by noise. It is proven that the covariance of the filtering error converges despite the nonlinear time-varying uncertainties in the system, and the filtering precision can be timely evaluated by the filter's parameters. Additionally, ESF is shown to be the linear minimum variance filter if the uncertainties are constant. The feasibility and effectiveness of ESF are illustrated by the simulation on a nonlinear spring-mass-damper system.