This article presents a method for structured robust control design for systems with a mixture of parametric and dynamic uncertainty. The proposed method alternates between an analysis step and a synthesis step. Samples of the parametric uncertainty are computed during the analysis steps, thus yielding an array of uncertain systems containing only dynamic uncertainty. The controller is then synthesized on this array of uncertain models. This synthesis step itself involves an alternation between constructing a D-scale for each of the uncertain systems and tuning a single controller for the entire collection of scaled plants. The controller tuning is performed using structured control design techniques. The proposed method is utilized to design a flutter suppression controller for a flexible aircraft. The aircraft dynamics are described by both a high-fidelity and a reduced-order model. The design objectives for flutter suppression are to achieve robust stabilization in the presence of mixed uncertainty. The proposed structured design method yields a single, low-order, linear time-invariant (LTI) controller, which increases the flutter speed by 15%. Additional robustness analyses and high-fidelity simulations are provided to assess the controller performance.

In the analysis of uncertain systems, we often search for a worst case perturbation that drives the $H_\infty $ gain of the system to the maximum over the set of allowable uncertainties. Employing the classical technique, an uncertainty sample is obtained, which, indeed, maximizes the gain but only at the single frequency where that maximum occurs. In contrast, this article considers a method to calculate a worst case perturbation that maximizes the gain of a system with mixed uncertainty at multiple frequencies simultaneously. This approach involves a nonlinear optimization that selects the worst case value of the uncertain parameters and the application of the boundary Nevanlinna–Pick interpolation to calculate the dynamic uncertainty sample. Such a perturbation can be used to augment Monte Carlo simulations of uncertain systems, especially if the system has multiple resonance frequencies. The worst case analysis of a flutter control system designed for a small flexible aircraft is provided to demonstrate the applicability of the proposed method.

The main direction of aircraft design today and in the future is to achieve more lightweight and higher aspect ratio airframes with the aim to improve performance and to reduce operating costs and harmful emissions. This promotes the development of flexible aircraft structures with enhanced aeroelastic behaviour. Increased aeroservoelastic (ASE) effects such as flutter can be addressed by active control technologies. Control design for flutter suppression heavily depends on the control surface sizing. Control surface sizing is traditionally done in an iterative process, in which the sizing is determined considering solely engineering rules and the control laws are designed afterwards. However, in the case of flexible vehicles, flexible dynamics and rigid body control surface sizing may become coupled. This coupling can make the iterative process lengthy and challenging. As a solution, a parametric control surface design approach can be applied, which includes limitations of control laws in the design process. For this a set of parametric models is derived in the early stage of the aircraft design. Therefore, the control surfaces can be optimized in a single step with the control design. The purpose of this paper is to describe as well as assess the developed control surface parameterized ASE models of the mini Multi Utility Technology Testbed (MUTT) flexible aircraft, designed at the University of Minnesota. The ASE model is constructed by integrating aerodynamics, structural dynamics and rigid body dynamics. In order to be utilized for control design, control oriented, low order linear parameter-varying (LPV) models are developed using the bottom-up modeling approach. Both grid- and polytopic parametric LPV models are obtained and assessed.

Various approaches to control aeroelastic wing section with structural nonlinearity emerged in the last years. One of the approaches is based on the Tensor Product (TP) model transformation and the feasibility test of linear matrix inequalities (LMIs) under the Parallel Distributed Compensation framework. The main contribution of this paper is that it extends the previously established qLPV model of the aeroelastic wing section with nonlinear friction. It is proven that the extended model can be exactly described after TP type polytopic decomposition. The qLPV model of the aeroelastic wing section with Coulomb friction was transformed to TP type polytopic model form for which LMI based controller was designed. The designed controller is compared to controllers of other authors dealing with the same problem.