EXPONENTIAL CONVERGENCE OF A NEW ERROR SYSTEM ARISING FROM ADAPTIVE OBSERVERS.

Abstract

The authors prove the uniform asymptotic stability of a novel error system, which has a transition matrix that contains both a regression vector and a filtered version of that vector. They show that stability occurs under conditions that are different from the usual SPR or small gain requirements. They also show that this error system arises from the design of adaptive observers for some classes of time-varying or nonlinear systems.