[BIC-announce] Special Seminar in Biomedical Engineering - Thursday June 20th - 10 am - Room 333

[Christophe Grova]

Dear all,

We will have a special Biomedical Engineering Dpt seminar this coming THURSDAY at 10h am .

This seminar is cosponsored by IEEE Engineering in Medicine and Biology (EMB), Montreal Chapter.


Thursday - June 20th,  at 10am

Location: Room 333 Lyman Duff Building (Biomedical Engineering Dpt, 3775  University Street).

Speaker: Dr. Michael Buchholz PhD, Ulm University, Germany

Title: Modelling of nonlinear systems using linear parameter-varying subspace-identification methods


Abstract:

Modelling of complex systems is always a compromise between high model accuracy and low model complexity. Hereby, the designated purpose of the model should always be taken into account. Besides first principle modelling, which leads to very detailed but in most cases also computational expensive models, system identification can be used to retrieve a mathematical description of the system behavior. For this identification, measurements of the inputs and outputs of the system are used to estimate the dynamical relationship between inputs and outputs, which in most cases leads to relatively simple models directly suitable for on-line applications like Kalman filtering, model-based control, or fault diagnosis.

For nonlinear systems, system identification without prior knowledge of the model structure is a challenging task. With linear parameter-varying (LPV) systems, a large number of nonlinear systems can be approximated very well. The LPV framework allows to describe a dependency of model parameters on specific signals, the so-called scheduling parameters, but retains the linear time-invariant (LTI) characteristics if these signals are constant. In this talk, an introduction to system identification based on subspace identification methods for linear parameter-varying systems will be given. After a short introduction to subspace methods for LTI systems, it will be shown how these methods can be extended to be applicable for bilinear and LPV system identification. To demonstrate the power of these methods, results from modelling of lithium-ion batteries for hybrid vehicles will be given, including the application of these methods for model order reduction to retrieve a on-board applicable simulation model for a non-measurable signal, which is required for optimized charging control.


A list of upcoming seminars can be found at : http://www.mcgill.ca/bme/news/seminars

See you there


Christophe Grova



***************************
Christophe Grova, PhD
Assistant Professor
Biomedical Engineering Dpt
Neurology and Neurosurgery Dpt

Multimodal Functional Imaging Lab (Multi FunkIm)
Montreal Neurological Institute
Centre de Recherches en Mathématiques

Biomedical Engineering Department - Room 304 McGill University
3775 University Street, Montreal, Quebec, Canada, H3A 2B4
email : christophe.grova at mcgill.ca
tel : (514) 398 2516
fax : (514) 398 7461

web:
http://www.bic.mni.mcgill.ca/ResearchLabsMFIL/PeopleChristophe
http://www.bmed.mcgill.ca/
MultiFunkIm Lab:
http://www.bic.mni.mcgill.ca/ResearchLabsMFIL/HomePage

***************************

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.bic.mni.mcgill.ca/pipermail/bic-announce/attachments/20130617/bc10958b/attachment.html>