A New Approach to Model Reduction Based on System Output Information Structure

Abstract A new approach to model reduction based on the utilization of the existing output information structure is proposed. The technique leads to a new system representation called the Generalized Hessenberg Representation (GHR). A numerical algorithm for obtaining the GHR is given. An example illustrates the use of the GHR for model reduction. The development and application of a generalized QL algorithm for model reduction based on the GHR are given. Its usefulness for model reduction is shown by an illustrative example.