Authors: M. Rewienski and J.K. White
Affilation: Massachusetts Institute of Technology,, United States
Pages: 128 - 131
Keywords: model order reduction, nonlinear dynamical systems, Krylov methods, piecewise-linear approach
In this paper we present an improved algorithm for nonlinear model order reduction (MOR) based on the trajectory piecewise-linear method proposed in  which, as shown before, provides an e ective and eÆcient strategy for automatically generating low-cost macromodels of highly nonlinear dynamical systems. The proposed extension consists of applying a more sophisticated projection basis, which merges multiple reduced order bases in Krylov subspaces generated at different linearization points. As a result we obtain a 'richer', aggregated reduced basis which, as shown on the example of a micromachined switch, enables us to improve the accuracy and further reduce the order of the macromodels for the considered highly nonlinear device.