Multimoment Matching Analysis of One-Sided Krylov Subspace Model Order Reduction for Nonlinear and Bilinear Systems

Modelling of complex systems comes with higher complexity and computational costs. Model order reduction enables better models to be exploited by control, diagnosis and prognosis algorithms. Effective model order reduction requires efficient methods to generate reduced order models. This is where the use of Krylov subspaces for model order reduction is of great advantage, especially when associated with high order bilinear model approximations of high order nonlinear models. This paper demonstrates the use of an Improved Phillips type projection for a one-sided Krylov subspace projection for reducing bilinear models. A new multimoment matching analysis of the proposed model reduction scheme is provided, and compared to some existing results in the literature.