Estimation of Bounded Model Uncertainties

Abstract

We identify parameters of a given input-output model so that estimated model output is consistent with the measured output of the system modeled. Parameter estimation based on a set-membership approach is a nonprobabilistic method for characterizing the uncertainty with which each model parameter is known. The model is consistent with data if the estimated output domain contains measured system output at each instant. Dynamic linear Multi-Input Multi-Output (MIMO) models are considered in this paper. Every equation error is bounded while model parameters fluctuate within a time-invariant domain represented by a zonotope. Our proposal helps find the characteristics of this domain, e.g., center, shape, size, by taking into account coupling between bounded variables of output equations to increase model accuracy.