ACS318 System Identification

Module Description

Modelling dynamical systems from first principles via Newton’s, Kirchoff’s or other known physical laws is often challenging and costly, requiring substantial expertise. An alternative is offered through “system identification” that takes observations of inputs and outputs from physical systems and infers or estimates a dynamical model directly.
This  module introduces two main ways of thinking about the identification problem, the theoretical framework that underpins them and the algorithms that compute the model estimates. It uses synthetic and real problems to illustrate the process and shows how models can be validated for future use.

Credits: 10 (Autumn semester)

Pre-requisites: ACS221 and ACS214

Module Leader

Professor V KadirkamanathanProfessor Visakan Kadirkamanathan

If you have any questions about the module please talk to me during the lectures or the labs in the first instance. It is likely that other students will learn from any questions you ask as well, so don’t be afraid to ask.

Outside of lectures please contact me via email, or drop by.

Learning Outcomes

Learning Outcomes

By the end of the module students will:

  1. Identify and classify the archetypal parametric and non-parametric paradigms of linear system identification and propose and select appropriate candidate solutions. [SM2m, EA1m, EA2m, EA4m]
  2. Explain the process of system identification and the requirements placed on the signals and systems involved. [SM2m, EA1m, EA4m, D2p, D3m, EP1m]
  3. Derive and solve key classes of system identification problem. [SM2m, SM5m]
  4. Explain and apply suitable theoretical tools for the evaluation of identified models. [SM3m, EA1m]
  5. Design and implement suitable experiments for system identification and interpret results in a critical way. [SM3m, SM5m, EA1m, EA2m, D3m]

This module satisfies the AHEP3 (Accreditation of Higher Education Programmes, Third Edition) Learning Outcomes that are listed in brackets after each learning outcome above. For further details on AHEP3 Learning Outcomes, see the downloads section of our accreditation webpage.

Syllabus

Syllabus

1. Introduction to system identification
2. Stochastic processes, correlation functions and power spectra
3. Nonparametric identification and pseudo random binary sequences
4. Introduction to parameter estimation and parametric models
5. Estimation theory and least squares algorithm
6. Unbiased estimation methods
7. Time series analysis and forecasting
8. Recursive estimation
9. Model validation, model selection
10. MIMO and nonlinear systems identification

Teaching Methods

Learning and Teaching Methods

18 hours of lectures (9 x 2 hour lecture slots)
+ 2 hours tutorials
+ 2 x 2 hour labs = 4 hours of laboratory

Teaching Materials

Learning and Teaching Materials

All teaching materials will be available via MOLE.

Assessment

Assessment

One two hour written examination
No resit examination is available for this module.

Students are given a formative (unassessed) problem class in week 6 in which they will attempt Matlab demonstrations with full code. Students are encouraged to repeat the formative assessment individually.

Feedback

Feedback

You will have an opportunity to view marked Exam Scripts once Exam results have been confirmed by the Faculty and released to students. The date of this Review session will be announced by the Departmental Office.

You will be able to look at the exam paper and a sample solution on the MOLE page for this module. The paper and solution will be available after the exam period.

Student Evaluation

Student Evaluation

Students are encouraged to provide feedback during the module direct to the lecturer. Students will also have the opportunity to provide formal feedback via the Faculty of Engineering Student Evaluation Survey at the end of the module.

Recommended Reading

Recommended Reading

• Rolf Iserman and Marco Munchhof, “Identification of Dynamic Systems”, Springer, 2011.
• Lennart J Ljung, “System Identification: Theory for the User”, Prentice-Hall (2nd ed), 1999.
• Stephen A Billings, “Nonlinear System Identification: NARMAX Methods in the Time, Frequency, and Spatio-Temporal Domains”, Wiley, 2013.