This dissertation deals with Krylov-based model order reduction of large-scale linear dynamical systems. In particular, it focuses on the time-domain properties of a reduced system obtained by these methods, and on the choice of the single expansion point about which the moment matching has to be achieved. By proving the equivalence of moment matching and Laguerre-based order reduction both in time and frequency-domain, a time-domain interpretation for moment matching is offered. In addition, the problem of the choice of the expansion point is reformulated as finding the best choice for the free parameter α in the Laguerre basis. Accordingly, two methods for the choice of this parameter are presented based on minimizing two objective functions involving the Laguerre coefficients of the impulse response. These methods are then integrated in the Krylov-based approach resulting in computationally efficient order reduction algorithms.
«
This dissertation deals with Krylov-based model order reduction of large-scale linear dynamical systems. In particular, it focuses on the time-domain properties of a reduced system obtained by these methods, and on the choice of the single expansion point about which the moment matching has to be achieved. By proving the equivalence of moment matching and Laguerre-based order reduction both in time and frequency-domain, a time-domain interpretation for moment matching is offered. In addition, t...
»