Second-Order Model Reduction Based on Gramians

Abstract EN

Some new and simple Gramian-based model order reduction algorithms are presented on second-order linear dynamical systems, namely, SVD methods.

Compared to existing Gramian-based algorithms, that is, balanced truncation methods, they are competitive and more favorable for large-scale systems.

Numerical examples show the validity of the algorithms.

Error bounds on error systems are discussed.

Some observations are given on structures of Gramians of second order linear systems.