LPV System Identification of a Flexible Manipulator: A Grey-Box Approach
Description
In this presentation, Mercère will introduce a new approach for identifying the dynamical model of flexible manipulators. The structure of the identified model, chosen as a descriptor LPV model, is derived from the original, non-linear equations governing the behavior of the system. A set of experiments around different configurations is involved, which is suitable for an accurate measurement of the tip of the manipulator by video camera. The final estimation step is global, allowing the direct identification of the global model based on the collection of local experimental data.
One of the shortcomings of the developed approach, however, is the use of genetic algorithm for the minimization of the identification criterion. This technique has been chosen because of the complexity of the identification cost function. As part of the talk, Mercère will also discuss the initialization of non-convex optimization algorithms, paying specific attention to linear, time-invariant grey-box models. The talk will also include a discussion about an extension to LPV structure.
Date of Event
December 8, 2011 2:15 - 3:15 PM
Location
Alvin Sherman Library, Room 2053, 3301 College Ave., Fort Lauderdale (main campus)
NSU News Release Link
http://nsunews.nova.edu/mathematics-colloquium-series-closes-semester-talk-flexible-manipulators-dec-8/
LPV System Identification of a Flexible Manipulator: A Grey-Box Approach
Alvin Sherman Library, Room 2053, 3301 College Ave., Fort Lauderdale (main campus)
In this presentation, Mercère will introduce a new approach for identifying the dynamical model of flexible manipulators. The structure of the identified model, chosen as a descriptor LPV model, is derived from the original, non-linear equations governing the behavior of the system. A set of experiments around different configurations is involved, which is suitable for an accurate measurement of the tip of the manipulator by video camera. The final estimation step is global, allowing the direct identification of the global model based on the collection of local experimental data.
One of the shortcomings of the developed approach, however, is the use of genetic algorithm for the minimization of the identification criterion. This technique has been chosen because of the complexity of the identification cost function. As part of the talk, Mercère will also discuss the initialization of non-convex optimization algorithms, paying specific attention to linear, time-invariant grey-box models. The talk will also include a discussion about an extension to LPV structure.
https://nsuworks.nova.edu/mathematics_colloquium/ay_2011-2012/events/9
Presenter Bio
Guillaume Mercere has a Ph.D. and is an Associate Professor at University of Poitiers