Johan Schoukens November 2015
THREE PRESENTATIONS ON SYSTEM IDENTIFICATION
MONDAY 9th AND TUESDAY 10th NOVEMBER
Professor Johan Schoukens, Free University of Brussels and Visiting Professor, University of Warwick
Earlier this year, Professor Johan Schoukens was awarded the higher doctorate degree of D.Sc. from Warwick for his distinguished work over many years in the area of System Identification, in which he is one of the World’s leading authorities. System Identification deals with how to find out as much as possible of the behaviour of a system from input signal(s) to the system and output signal(s) from the system, combined with any prior knowledge of the system. The system could be of any type, electrical, mechanical, biological, etc. We are very fortunate that Johan is visiting the School of Engineering on 9th and 10th November to give three presentations.
Presentation 1: Why do you need system identification?
Monday 9th November, 10.00 to 12 noon, Room D202 (new extension-2nd floor), to be followed by a Q&A session.
It will be shown that system identification is needed whenever mathematical models are extracted from experimental data. It will be illustrated, by a simple example, which shows that following intuitive approaches can lead to poor models without giving any warning to the user. For that reason a systematic approach is needed.
Presentation 2: Identification of (linear) dynamical systems: A case study
Monday 9th November, 16.00 to 18.00, Room D202.
In this presentation, we illustrate the potential benefits of a sound system identification approach to data driven modelling. We first explain how to identify a parametric model for linear dynamical systems from experimental data. We start from a frequency domain approach, and give a MATLAB illustration, using experimental mechanical data and a freely available toolbox: FDIDENT. Next we discuss the impact of nonlinear distortions on this approach.
Presentation 3: A statistical framework for system identification
Tuesday 10th November, 14.00 to 16.00, Room D202, to be followed by a Q&A session in Room A401.
In this presentation, a systematic approach to the system identification problem is presented. We give a basic introduction to least squares, weighted least squares, maximum likelihood, and Bayesian estimation. We will also explain how to deal with disturbances on both the input- and output measurements. The theory will be illustrated with many examples.