Shift parameter sensitivity on the computation of unstable reduced order model via truncated balanced method

This paper presents reduced order models (ROMs) of unstable systems. The effect of a shift parameter is studied to stabilize the system and then apply the truncated balanced realization (TBR) approach to compute the reduced ROM. The extended Krylov subspace (EKS) technique is applied in order to obtain the ROM. To evaluate the performance of the methodology, experiments in a 606-state system model were assessed. The study demonstrated that the conditioning of the state matrix through a shifting of the eigenvalues works on properly for determination of unstable system ROM.