Auto-Tuned Multi-Objective Structured H2/H-infinity Control Design Methodology for Gust Load Alleviation

Kurzfassung

Aircraft structures are sized, in part, by gust and turbulence loads. Gust load alleviation (GLA) seeks to improve aircraft efficiency by reducing these loads through active control. However, their dynamic and unpredictable nature as well as the limitations imposed by system delays and dynamics make for a challenging control problem. Doppler wind lidar sensors allow atmospheric disturbances to be detected with ample lead time, significantly improving achievable GLA performance. Due to relatively high levels of measurement noise, such sensors require an estimator which produces a wind profile along the aircraft's direction of travel. A lidar-based GLA control design methodology must aim to make full use of this information while respecting system limitations, addressing secondary objectives, avoiding excessive actuator use, and ensuring an adequate level of robustness, not to mention questions of certifiability. Structured H-infinity optimal preview control has proven, in many respects, to be suitable for this purpose, however it has several drawbacks. Firstly, the certification specifications partly define gust loads in terms of time-domain peak responses to discrete gusts. The link between a system's transitory time-domain response and its H-infinity norm is indirect and often poor, so tuning H-infinity specifications to meet the time-domain requirements is seldom trivial. Secondly, the lidar-based wind estimation system necessarily introduces dynamic effects and noise-induced uncertainties into the wind estimate. These have heretofore been neglected in control synthesis, forcing the designer to indirectly compensate for them by iteratively adjusting the control specifications. Together, these issues result in a difficult and unnecessarily time-consuming design process, potentially degrading its achievable performance and limiting the degree to which it can be scaled up to industrial problems. This thesis aims to address these obstacles by developing and expanding the structured optimal preview control methodology. A new type of discrete gust specification based on the H2 norm and a novel discrete gust impulse filter is introduced, allowing the discrete gust response to be systematically tuned. A method for analytically computing a linear model of the combined lidar and wind estimation system is developed, and the resulting model is added to the control problem, ensuring that control synthesis takes its characteristics into account. Finally, on the basis of the previous two developments, an automated iterative control tuning scheme is proposed and developed, largely eliminating the need for manually adjusted control specifications. Though not demonstrated in the present work, this method opens the door to integrated aeroservoelastic control design in multidisciplinary aircraft design