Dissipativity-Preserving Model Reduction Based on Generalized Singular Perturbation

In this paper, we develop a dissipativity-preserving model reduction method based on a generalized singular perturbation approximation. This model reduction framework can deal with not only standard singular perturbation approximation but also projection-based model reduction as a special case. To develop such a model reduction method, we investigate a condition under which system dissipativity is appropriately preserved through the approximation. Moreover, deriving a novel factorization of the error system in the Laplace domain, we derive an a priori error bound in terms of the H-2-norm. The efficiency of the model reduction is shown through an example of interconnected second-order systems.