Identifying robust subspaces for dynamically consistent reduced-order models

Document Type

Conference Proceeding

Date of Original Version



Nonlinear models for large/complex structures are hard to simulate for parametric studies of long-time dynamical behaviors. Reduced order models (ROMs) provide tools that both allow long-time dynamical simulations and reduce data storage requirements. In data-based model reduction, there is usually no consistent way to determine if the obtained ROM is robust to the variations in system parameters. Here, we use two concepts of "dynamical consistency" and "subspace robustness" to evaluate ROMs validity for parametric studies. The application of these concepts is demonstrated by reducing a nonlinear finite element model of a cantilever beam in a two well potential field. The resulting procedure can be used for developing and evaluating ROMs that are robust to the variations in the parameters or operating conditions. © The Society for Experimental Mechanics, Inc. 2014.

Publication Title

Conference Proceedings of the Society for Experimental Mechanics Series