Meixner-like parameters identification using interval analysis

Abstract

In this paper, the Meixner-like model is used to represent the linear discrete-time system. Furthermore, we present, from input/output measurements a recursive representation of Meixner-like model. The minimization of the Normalized Mean Square Error is considered to estimate the optimal Meixner-like pole. The belonging domain of the parameters, which is compatible with the model structure, measurements and the bounds of the error, is defined by the interval values. A numerical simulation shows the efficiency of the approach.