Krylov subspace method for passive reduced-order modelling of electrical networks

This study presents a two-stage method to compute the passive reduced-order models of multi-input multi-output electrical networks containing distributed-parameter transmission lines (TLs). The method is based on the discretised TL models and a Krylov subspace model-order reduction method called passive reduced-order interconnect macromodelling algorithm. The section numbers of the TLs are adaptively determined using an innovative and fast algorithm. In stage 1, the high-order cascaded Pi circuit models of the TLs are reduced and synthesised into low-order RLC two ports; stage 2 further reduces the entire lumped network. System passivity is preserved during the reduction. Useful postprocesses and a practical passivity test technique are given. Numerical tests confirm the superiority of the proposed method in reliable passivity preservation with decent accuracy and efficiency.