Reduced-order modelling of large linear passive circuits using Krylov projection method

Abstract

The efficient simulation of large-scale dynamic systems needs a systematic procedure for order reduction of the original circuit. The paper combines the projection method with Krylov subspaces to obtain robust and accurate order reduction techniques. First of all, the algorithm finds a basis matrix onto Krylov subspaces, using a block Arnoldi process, and then constructs a reduced-order model via projection technique. It is shown that we can handle the MIMO systems using the semi-state method and Krylov projection method. An illustrative example is given and some conclusions are pointed out.