Identification of continuous-time parametric models

System identification involves the search for a mathematical model of a dynamic system, based on experimental data and knowledge available "a priori". This macroscopic model is characterized by a structure and by parameters that need to be chosen and adjusted, in order to best reproduce the system's input-output behavior.

Traditionally, system identification methods are used to determine models for the synthesis of control laws. However, this conventional field of application is not the only one, and the methods are also used to :

• estimation of physical parameters that cannot be directly measured;

• model-based system diagnostics ;

• simulation, used for design, forecasting or training purposes;

• test interpretation.

Although the methods are mainly developed by automaticians and applied mathematicians, they can be used in a wide variety of fields, from manufacturing processes to econometrics, biology, transport or environmental processes.

The aim of this dossier is to raise awareness of continuous-time model identification methods, whose algorithms are now also available in software libraries , to give an overview of recent developments, and to present some application results for these methods.