6
System Identification Identification de syst` emes dynamiques Alireza Karimi Laboratoire d’Automatique ME-C2 397 email:[email protected] Site: http://la.epfl.ch, Teaching Fall 2013 (Introduction) System Identification Fall 2013 1/6

System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

System Identification

Identification de systemes dynamiques

Alireza KarimiLaboratoire d’Automatique

ME-C2 397 email:[email protected]

Site: http://la.epfl.ch, Teaching

Fall 2013

(Introduction) System Identification Fall 2013 1 / 6

Page 2: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

Content

1. Plants, systems and models (5h c + 3h Lab )

Modeling, Type of modelsRepresentation methods

2. Nonparametric models (9h c + 6h Lab)

Time-domain methods (step and impulse response,correlation method)Frequency-domain methods (Fourier analysis, Spectralanalysis)Closed-loop identification

3. Parametric models (13h c + 6h Lab)

Linear regression (Least squares method, recursivemethods)Prediction error methodsPractical aspects (Order estimation, validation)

(Introduction) System Identification Fall 2013 2 / 6

Page 3: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

Course Schedule

System Identification, Fall 2013Tuesday 8:15 – 11:00

Cours: BM5202 and Lab:ME A0 392Introduction

17 Sep. ModelingRepresentation methodsContinuous- and discrete-time models

24 Sep. Choice of sampling periodStep and impulse response analysis

1 Oct. IdLab1: Step and Impulse response

Auto-correlation, Cross-correlation, Random signals8 Oct. Correlation method

Excitation signal PRBSFrequency-domain methods

15 Oct. Truncation errorsMatlab demo

22 Oct. IdLab2: Correlation, PRBS

(Introduction) System Identification Fall 2013 3 / 6

Page 4: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

Course Schedule

Spectral analysis29 Oct. Closed-loop identification

Parametric models from frequency dataLeast squares method

6 nov. Recursive least squares methodBias and variance error, Instrumental variables

12 Nov. LdLab3: Frequency-domain methods

Prediction error method19 Nov. ARX, ARMAX, OE, BJ structures

Bias and variance analysisPractical aspects

26 Nov. Order estimationValidation

3 Dec. IdLab4: Parametric Identification (CO5-CO6)

Identification en boucle fermee10 Dec. Identification Toolbox Demo

Project definition

17 Dec. IdLab5: Final Project

(Introduction) System Identification Fall 2013 4 / 6

Page 5: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

Identification Lab

Objectives:

Practice the identificationalgorithms in simulation.

Become familiar with theIdentification Toolbox of Matlab.

Parametric identification andvalidation of the model of amechanical system.

Complete system identification(parametric and non-parametric)and validation using the realmeasured data of a

3DOF Gyroscope

3 DOF Gyroscope

(Introduction) System Identification Fall 2013 5 / 6

Page 6: System Identification · mechanical system. Complete system identification (parametric and non-parametric) and validation using the real measured data of a 3DOF Gyroscope 3 DOF

Course notes and Exam

Evaluation:

Brief report for each Lab (20% of the final grade).

Brief report for the final project (20% of the final grade).

Oral Exam:

1 Each student chooses 2 questions from a list of 40-50 questions.

2 He/She prepares the answer during 20 minutes.

3 He/She answers to the chosen questions (40% of the final grade).

4 He/She answers to the questions about the final project (20% of the finalgrade).

Course notes (in French):

D. Bonvin, A, Karimi, Polycopie : “Identification de systemes dynamiques”,Edition Septembre 2011.

(Introduction) System Identification Fall 2013 6 / 6