Proper orthogonal decomposition versus Krylov subspace methods in reduced-order energy-converter models

In this paper, the proper orthogonal decomposition and the Arnoldi-based Krylov subspace methods are applied to the magnetodynamic finite element analysis of power electronic converters. The performance of these two model order reduction techniques is compared both in frequency and time domain. Moreover, two original, adaptive and automated greedy snapshots selection methods are investigated using either local or global quantities for selecting the snapshots (frequencies or time steps).