Aerospace vehicle model simplification for feedback control
This dissertation addresses three distinct topics within the broader scope of aerospace vehicle model simplification for feedback control. These topics include order reduction criteria for feedback control, frequency weighted internally balanced order reduction, and symbolic factoring of transfer function polynomials. Before ever reducing the dynamic order of a model located within a feedback system, it is critical to understand which dynamic characteristics are crucial to preserve so that the reduced order closed-loop system will exhibit those key characteristics similar to the higher order closed-loop system. Based upon closed-loop stability and stability robustness requirements, order reduction criteria are derived, specifically for sequential loop closure settings. The weighted, internally balanced truncation technique delivers a reduced model that is accurate in a frequency range specified by the user. As yet, however, no existing a priori error bound has been obtained for the technique. Building upon a previous attempt, an error analysis is presented which gives support to the technique. Also, a new reduction technique, which combines weighted, balanced coordinates with residualization, is presented. The functional dependence of linear system poles and zeros upon the basic system parameters appearing in the governing differential equations offers insight into the causes for key dynamic characteristics. A Taylor series based approach is offered to generate approximate symbolic expressions for the poles and zeros in terms of the basic system parameters. ^
Major Professors: David K. Schmidt, Purdue University, Dominick Andrisani II, Purdue University.