COMBINING CHARACTERISTIC NUMBERS TO REDUCE THE UNCERTAINTY OF FRACTIONAL-ORDER PROCESS MODELS

SCHLEGEL, M.; ČECH, M.

Abstract

The paper extends the authors' previous results where the class of all lag/dead time process models (with integer or fractional order poles) is considered in the case that the following characteristic numbers are given: a) one sample of a process frequency response, b) first three moments of the process impulse response. In both cases the explicit parameterisations of the value set boundary were provided. In this paper, the simultaneous knowledge of the characteristic numbers a) and b) is assumed and the corresponding upper hard bound of the value set boundary is described. It is shown, that combining experimental data a) and b) leads to the significant reduction of a model uncertainty in the frequency domain. Additionally the investigated results are applied to the robust controller design.

Coresponding author e-mail: mcech[at]kky[dot]zcu[dot]cz

Session: Modelling, Simulation, and Identification of Processes