Seminar Details
2010-09-14 (14.00) : Refined instrumental variable methods for identification of Linear Parameter-Varying (LPV) Box-Jenkins models
At Euler building
Organized by Mathematical Engineering
Speaker :
Vincent LAURAIN (CRAN, Nancy-University, CNRS, France)
Abstract :
The common need for accurate and efficient control of today's industrial applications or modelling of natural systems is driving the system
identification field to face the constant challenge of providing better models of physical phenomena. This rises the need for system
descriptions that form an intermediate step between Linear Time-Invariant (LTI) systems and nonlinear/time-varying plants.
To cope with these expectations, the model class of Linear Parameter-Varying (LPV) systems provides an attractive candidate.
In LPV systems the signal relations are considered to be linear just as in the LTI case, but the parameters are assumed to be
functions of a measurable time-varying signal, the so-called scheduling variable. This LPV modelling concept allows for a wide
representation capability of physical processes but the real practical significance of the LPV framework until now lays in its
well worked out and industrially reputed control synthesis approaches that have led to many successful applications of LPV control
in practice. Despite the advances in the LPV control field, identification of such systems is not yet well developed. Most methods
assume static dependence on the scheduling parameters and it can be shown that it is generally impossible to reach statistically
optimal estimates by using linear regression as presented so far in the literature. Furthermore, there was a lack of an LPV identification
approach which is capable of efficiently estimating LPV-IO models underrealistic colored noise conditions.
Moreover, LPV identification methods are almost exclusively developed for DT as in this setting it is much easier to handle the estimation of
parameter-varying dynamics but there is a growing need of the LPV framework for efficient identification methods that directly deliver
reliable CT models.
In this seminar, a short horizon on available methods from the literature for identifying LPV IO models will be first given. An
instrumental variable method for identifying discrete-time LPV IO models under colored noise conditions will be presented along with its
extension to handle the direct identification of continuous-time LPV models.