Entry requirements

Either Spectral analysis, Signals and systems or Stochastic modelling.

Aims

To give knowledge about computerised

methods and computer aided tools for

calibrating parameters in dynamic models using

measurement data (system identification).

To give knowledge on how to analyse system identification methods using a statistical approach.

Content

The system identification problem - from data to model. The least

squares method. Model structures and input

signals. The prediction error methods.

The instrumental variable methods. Model

validation and practical aspects. Identification of feedback systems.

Laboratory work: Five two-hour MATLAB

laboratories:

1. The System Identification Toolbox.

Introductory examples

2. Prediction error methods and instrumental

variable methods.

3. Recursive identification.

4. Practical aspects.

5. One half-day laboratory: Identification and

control of a laboratory process.

5. Time series analysis.

One half-day laboratory: Identification and

control of a laboratory process.

Instruction

Lectures, Problem solving sessions, Laboratory work and homework assignments.

Assessment

Written examination at the end of the course.

Passed laboratory course is also required.