A variable structure parallel observer system for robust state estimation of multirate systems with noise

Abstract

This paper proposes the variable structure parallel observer system for noisy measurements (VSPOSN), a multirate estimation technique for discrete time systems, to compensate for output measurements that are noisy and only periodically available. The VSPOSN is an extension of the original Parallel Observer System. The VSPOSN is based upon a stable and robust discrete-time sliding mode observer with a set of attractive boundary layers driving a Luenberger observer running in parallel. The resulting observer system provides stable state estimates for each control step despite incomplete and noisy measurements. The VSPOSN is applied to a second order resonance model with nonlinear uncertainties/disturbances as proof of concept, and a stability proof and error bounds are provided.