Identification of ARMAX systems with bounded noise

Abstract

This paper focuses on the identification of ARMAX systems in the context of unknown but bounded noise. Two identification algorithms are proposed and analyzed. These algorithms are Ellipsoidal Outer Bounding (EOB) algorithms. The first algorithm is a recursive algorithm. The analysis of this first algorithm shows that it has a limited implementation framework. The second algorithm overcomes some limitations of the first algorithm. The implementation of the second algorithm is realized using an iterative procedure. Numerical simulations are proposed to confirm the analysis and to show performance of the proposed approach.