Control Systems
The department Aeroelastic Control Systems researches novel functions and algorithms, as well as new underlying methods for control design and verification. Examples include functions to reduce gusts and maneuver loads, active flutter suppression, and structural damping to increase passenger comfort. Control design focuses on both the overall process (based on principles of systems engineering - from requirement specification to implementation and testing) and highly automated fast synthesis methods as part of the overarching aircraft design process. Multi-objective parameter optimization and modern control engineering methods form an important basis. We validate new functions and methods in wind tunnel tests, as well as on manned and unmanned flight test demonstrators.
The core of such functions is nearly always based on closed-loop control. This means that control commands are continuously calculated and applied to the aircraft based on the current aircraft and airframe state as registered by sensors (like accelerometers). Feedback elegantly allows the effective integration of individual components that may have uncertain behavior are may be subject to disturbances, into effective functions. The dynamics and equilibrium of a system (in this case, the aircraft structure) are positively influenced.
On the other hand, however, this feedback can quickly lead to unstable system behavior, excessively load system components, push actuators and sensors to their limits, or bring the entire system into dangerous states. If components fail, reconfiguration is often necessary. Finding suitable compromise solutions makes the design of control algorithms an exciting and challenging task.
For the actual design, the field of control engineering has an ever-growing range of methods and tools for evaluating and synthesizing dynamic systems. Some methods are generally valid, others are suitable for special classes of systems (for example, systems that record very nonlinear behavior).
Our department, which cooperates closely with neighboring departments at the Institute of Flight Systems Technology and the Institute of System Architectures in Aviation, is dedicated to the further development of control engineering methods and tools. We focus particularly on applications that improve the dynamic behavior of the aircraft from an aeroelastic point of view. Expected loads are a dimensioning factor in the design of the aircraft structure. With the help of control functions, these loads can be significantly reduced, which allows the structure to be dimensioned lighter. It is also very important that a high level of reliability and safe integration into the overall flight control system are guaranteed.
Dynamic models of the overall system play an important role in the design process. Even if individual synthesis methods do not always need them, the necessary verification using simulations or other model-based analyses form an important part of the approval process. Good and fast models contribute to significant cost reductions at this point.
Control engineering analysis methods often require special model representations, such as linear state forms, transfer functions, or linear fractional representations (LFRs). The control algorithms themselves are usually compiled graphically using so-called block diagrams in tools such as Matlab/Simulink. The blocks can come from a mathematical synthesis method for control laws or be parameterized filters, gains and time constants that can be adjusted using multi-objective optimization to best meet the above requirements.
Robust control
Robust control methods offer the most comprehensive extension of classical control theory (the development of which began at the beginning of the 20th century) to systems with multiple input variables and output variables. They make it possible to achieve the best possible performance despite given uncertainties in the system. In addition, the internal structure of the controller is inherently optimized. This is particularly useful for structural control functions.
Input-Output Blending
A recent example of method development is based on an efficient separation between feedback control, synthesis of the feedback signals, and control allocation. This so-called input-output blending method is particularly suitable for the damping of specific structural modes (for example potential flutter modes). The actual control problem is reduced to a minimum dimension by optimally combining sensor signals or control commands as seen and used by the controller. The combination of the sensor signals aims at the highest possible observability, the combination of the control deflections at a maximum effect on the critical modes.
Control with artificial intelligence
This is a rapidly growing area that uses artificial intelligence methods. It enables control algorithms or sub-functions to learn online or offline and thereby improve the dynamic behavior of the system. This is a long-term research, but the possibilities seem endless. We have already flight-tested the first methods, with promising results!
Optimization
We use a lot of optimization in our design processes, especially as a suitable tool to avoid manual trial and error when trying to meet many requirements at the same time.
Optimization strategies make it possible to find compromise solutions for complex problems. The validation and analysis of designed systems can also be formulated as an optimization task in which the design requirements are mapped into criteria and the so-called worst case for all permitted parameter variations is sought with the help of optimization.
An important aspect and main goal of our work is always the implementation and execution of tests in the wind tunnel or in flight test so that the practical suitability of new functions or design methods can be assessed. Later certifiability is also taken into account.