Course detail

Modelling and Identification

FEKT-NMIDAcad. year: 2011/2012

The subject is oriented on:
- identification methods of dynamic systems
- approaches towards nonparametric and parametric identification
- on-line and off-line identification
- spectral estimation, assessment of noise and disturbance influence on identification results

Language of instruction

English

Number of ECTS credits

6

Mode of study

Not applicable.

Guarantor

doc. Ing. Petr Blaha, Ph.D.

Department

Department of Control and Instrumentation (UAMT)

Learning outcomes of the course unit

Students acquire knowledge and experience in various methods of identification of dynamical systems especially with help of Matlab and its toolboxes.

Prerequisites

The subject knowledge on the Bachelor´s degree level is requested.

Co-requisites

Not applicable.

Planned learning activities and teaching methods

Teaching methods depend on the type of course unit as specified in the article 7 of BUT Rules for Studies and Examinations.

Assesment methods and criteria linked to learning outcomes

Numerical Exercises - Max 15 points.
Individual project - Max. 15 points.
Final Exam - Max. 70 points.

Course curriculum

1. Introduction into dynamic system identification
2. Nonparametric identification methods
3. Input signal for identification
4. Least squares method
5. Dynamic system models for system identification
6. Recursive LSM
7. Instrumental variable methods
8. Identification methods based on prediction error whitening
9. Practical notes on system identification
10. Identification using neural nets and fuzzy modeling
11. Another approaches to system identificaiton
12. Identification of nonlinear dynamic systems
13. Course summary

Work placements

Not applicable.

Aims

Familiarize students with basic techniques for dynamic system identification and with their possible limitations. The students will get to know how the noise acting on the plant influences the identification results and how to cope with it.

Specification of controlled education, way of implementation and compensation for absences

The content and forms of instruction in the evaluated course are specified by a regulation issued by the lecturer responsible for the course and updated for every academic year.

Recommended optional programme components

Not applicable.

Prerequisites and corequisites

Not applicable.

Basic literature

Ljung, L.: System Identification, Theory for the User, Prentice Hall, 1987 (EN)
Soderstrom T., Stoica P.: System Identification. Prentice Hall International, 1989 (EN)

Recommended reading

Not applicable.

Classification of course in study plans

  • Programme EECC-MN Master's

    branch MN-KAM , 2 year of study, winter semester, elective specialised
    branch MN-BEI , 1 year of study, winter semester, elective interdisciplinary

Type of course unit

 

Lecture

26 hod., optionally

Teacher / Lecturer

doc. Ing. Petr Blaha, Ph.D.

Syllabus

Introduction into problematic of dynamic system identification.
Nonparametric methods of identification.
Linear regression and least squares method.
Useful excitation signals, persistent excitation, pseudorandom binary sequence.
Prediction error method.
Instrumental variable method.
Recursive identification methods, numerically stable identification methods.
Spectral estimation, AR, MA and ARMA models.
Identification in closed loop.
Validation of identified model.
Kalman filter, extended Kalman filter.
Practical notes to identification.
Summary of acquired knowledges about identification of dynamic systems.

Exercise in computer lab

26 hod., compulsory

Teacher / Lecturer

doc. Ing. Petr Blaha, Ph.D.

Syllabus

Stochastic signals and their statistical evaluation.
Basic methods for nonparametric identification.
Least squares error method.
Generation of excitation signals.
Recursive least squares method.
Influence of noise acting in different places of the system on identification results.
Basic commands from MATLAB Identification Toolbox.
Utilization of MATLAB Identification Toolbox.
Utilization of MATLAB Identification Toolbox.
Spectral estimation of discrete time models.
Experiments with Kalman filter.
Quality evaluation of the identified method.
Exercises evaluation.