ΕΚΔΗΛΩΣΕΙΣ


Τύπος: Διάλεξη
Ημ/νία Έναρξης: 13/1/2010
Ημ/νία Λήξης:
Ώρα Έναρξης: 13:00
Ώρα Λήξης:
Ομιλητής: Ελένη Ν. Χατζή, Υποψήφια Διδάκτωρ, Department of Civil Engineering and Engineering Mechanics, Columbia University, New York, ΗΠΑ
Τίτλος: Application of On-Line Parametric Identification for Nonlinear Systems
Περιγραφή: Over the past years, there has been interest in the efficient simulation and identification of nonlinear structural system behavior. This presentation will explore the application of the most recent identification techniques in two classes of problems. The first one is a comparative study of the performance of the Unscented Kalman Filter (UKF) and Particle Filter methods (also known as Sequential Monte Carlo method (SMC)) on employing accelerometer and displacement sensor measurements for structural system identification. In particular, the use of GPS displacement measurements coupled with accelerometer measurements from different locations is explored. Higher order non linearities call for the application of advanced methods as opposed to the well known Extended Kalman Filter (EKF), which is only reliable for systems that are almost linear on the time scale of the updating intervals. The efficiency of the aforementioned techniques is evaluated through the example of a three dof system, with a Bouc-Wen hysteretic component. The second problem deals with the on-line identification of non linear hysteretic systems where not only the parameters of the system are unknown but also the nature of the analytical model describing the system is not clearly established. To this end a Bayesian approach using the UKF method has been applied in order to investigate the effects of model complexity and parametrization. The state space formulation incorporates a Bouc-Wen type hysteretic model properly modified with additional polynomial or exponential-type nonlinear terms that are properly weighted throughout the identification procedure. In addition, a twofold criterion based on the smoothness of the parameter prediction and the accuracy of the estimation is introduced in order to investigate the required model complexity as well as to potentially rule out ineffective terms during the identification procedure (on-line). The method is validated through the identification of the nonlinear hysteretic behavior produced by an experimental setup, designed by Tasbihgoo and Masri, involving displacement and strain (restoring force) sensor readings.
Χώρος: Αμφιθέατρο
Link1:
Link2:
Photo1:
Photo2:
Photo3:

Επιστροφή