On model updating of existing structures utilizing measured dynamic responses

Abstract

Life-cycle performance evaluation of structures relies heavily on the usability of measured data. Typically, measurements are utilized to update the state of knowledge about a structure. This usually leads to a system identification problem. Due to the ill-conditionedness of inverse problems, the actual value of the obtained data may be severely reduced. Selective sensitivity provides a methodology to improve the condition of the inverse problems by transforming the original (large) problem into a sequence of smaller ones. The paper discusses this concept and develops a novel alternative concept for the development of selectively sensitive excitations, which does not require precise knowledge of the system parameters.

Keywords:

• Structural dynamics
• System identification
• Model updating
• Selective sensitivity
• Bayesian methods

Acknowledgements

This work has been carried out within the Collaborative Research Center SFB 524 supported by the German Research Council (DFG). The first author developed significant parts of the concept as presented during sabbatical stays at the University of Tokyo (with Professor Tsuyoshi Takada) and the University of Colorado at Boulder (with Professor Dan M. Frangopol). The second author has been supported by a scholarship of the German Academic Exchange Service (DAAD). All support is gratefully acknowledged.

Notes

First published online 2 March 2005.

Additional information

Notes on contributors

Christian Bucher

First published online 2 March 2005.