Lecturer: MER, Dr. Alireza KARIMI, ME C2 397, tél: 33925.
Some methods are studied for identification of discrete-time linear models using experimental data. The correlation method and spectral analysis are used to identify nonparametric models and the prediction error method to estimate the plant and noise model parameters. Hands-on labs are included.
- Modelling, type of models and representations.
- Time-domain nonparametric identification methods (impulse response by the correlation aproach).
- Frequency-domain nonparametric identification methods based on the Fourier and spectral analysis.
- Parametric identification by linear regression (least squares method, instrumental variables method, recursive algorithms).
- Introduction to subspace identification.
- Prediction error methods (ARX, ARMAX, OE and BJ structures).
- Practical aspects of identification (input design, order estimation, model validation).
- Plant model identification in closed-loop operation.
By the end of the course, the student must be able to:
- Identify a dynamic system using experimental data,
- Construct and analyze a discrete-time model for a dynamic system,
- Examine the performance and the solutions and draw conclusions.
Document: Cours-notes “System Identification” by Alireza Karimi
Prequisites: Control Systems.
Important concepts to start the course:
- Represent a physical process as a system with its input, outputs and disturbances
- Analyze a linear dynamical system (both time and frequency response)
- Represent a linear system by a transfer function (discrete- and continuous-time)
Teaching method: Ex-cathedra course with computer-based exercises and project
Information on computer exercises and project: See Moodle.