A Singular Perturbation Model Order Reduction Method for a Specially Structured Model

In this paper we consider a model order reduction technique in the context of a special descriptor representation of linear time-invariant continuous-time systems. Specifically, we present a unique form of singular perturbation model order reduction technique for such a representation. It is shown that this technique leads to a reduced-order model with an upper triangular system matrix. This method also results in one algebraic Riccati equation instead of two, which are generally required for singular perturbation model order reduction techniques. The foregoing structure enables the analysis and design of large dimensional systems with less computational cost and errors in comparison to the existing methods. An example is also given to illustrate the application of the proposed method.