Characterizing model variation for robust control of flexible atmospheric flight vehicles

The aerospace community has shown significant interest in the next generation of large high-speed atmospheric flight vehicles such as the High-Speed Civil Transport (HSCT) and hypersonic vehicles similar to NASP. This class of aircraft is characterized by low structural stiffness and significant aeroelastic interactions.; Typically, controllers are designed and analyzed based on a nominal finite-dimensional linear time invariant system model to meet desired aircraft flight performance requirements and closed-loop stability. However, the real dynamic system varies from the mathematically modeled system due to linearization, unmodeled dynamics, and variations in flight condition, for example. Furthermore, the frequencies and mode shapes of the aeroelastic modes are not well variations are usually treated through the use of an “uncertainty” or variation model.; The many advances in multivariable robustness theory such as μ analysis offer the promise of powerful tools for evaluating the robustness of multivariable feedback systems. However, an important assumption is that a variation model is available, prior to robustness analysis, which adequately characterizes the variation known. These in the nominal system design model.; To be presented and discussed is the development of a systems approach to generating a variation model characterizing the variation in finite dimensional linear time-invariant models of flexible aircraft arising from unsteady aerodynamic effects, structural mode truncation, and uncertainty in the mass and stiffness properties of the aircraft structure. The approach takes into account the dependant nature of structural mode frequencies and structural mode shapes on the fundamental mass and stiffness properties of the aircraft structure.; Application of the systems approach is then demonstrated by generating a variation model for an example aircraft representative of the class of large flexible aircraft. Using the generated variation model it is shown that model variation resulting from unsteady aerodynamic effects can be significant in frequency regions where active structural mode control is desired.; It is also shown that current approaches to specifying model variation underestimate the level of model variation in two significant ways. First, by not properly accounting for the aerodynamic coupling between the rigid body and structural modes and second, by not properly accounting for model parameter uncertainty. In underestimating model variation, robustness and performance properties of control designs, based on incorrect assumptions of model variation, are questionable.