English

If sufficient demand

System identification treats how to construct mathematical models of dynamical systems from measured data. The doctoral student acquires knowledge about the ideas, concepts and theory of system identification. The doctoral student develops the ability to preform system identification experiments and to perform system identification using experimental data.

The mathematical foundations of System Identification. Parametric and non-parametric techniques. Parametrizations and model structures. Parameter estimation.
Asymptotic statistical theory. User choices. Experimental design. Choice of model structure. Assessment of the results.

know and understand the fundamental ideas of system identification and the theoretical framework.

have the ability to plan and execute system identification experiments, choice of input signals and selection of appropriate models and methods.

demonstrate skills in using the "System Identification Toolbox" and the ability to analyze and evaluate the results of the calculations.

have the ability to judge the appropriate model complexity required for control design.

have the ability to balance model complexity and efforts required for experiments and analysis.

Lectures

Seminars

Exercises

Project

The instruction is a mixture of lectures and seminars. Problems are assigned every week and discussed the next week following the schedule for Ljungs course at ISY in Linköping http://www.rt.isy.liu.se/student/graduate/idkurs/. A project has also to be executed preferably related to the students PhD subject. The doctoral students can collaborate in the project.

Oral exam

Written assignments

Evaluation of the weekly problem solving session. An assessment of the project and an oral examination.

Failed, pass

Linear systems, probability theory and statistics

Ljung, L.: System Identification - Theory for the user 2n edition,. Prentice Hall, 1999.

Also Mathworks: System Identification Toolbox.

FRT115F

 -02-10

FN1/Anders Gustafsson