A projection-based algorithm for model-order reduction with H2 performance: A convex-optimization setting

A suboptimal and a simple solution to the H2-model-order-reduction problem is given. The proposed procedure is a combination of projection and H2 Model-order Reduction (MOR) techniques. To show the efficacy of the proposed method, a set of comparisons with classical MOR methods is provided with numerical simulations. The numerical algorithm is subsequently employed in the design of reduced-order nonlinear systems satisfying the Input-to-State Stability (ISS) with respect to their inputs. Reduced-order models for nonlinear electrical circuits are successfully built to demonstrate the usefulness and the performance of the developed procedure.