Flight Test with an Adaptive Feed-Forward Controller for Alleviation of Turbulence Excited Wing Bending Vibrations

Abstract

On large, and thus highly flexible airliners, atmospheric turbulence excited wing bending vibrations generally cause high dynamic loads in the wing roots and degrade the ride comfort. In the past, these problems have been tackled by the use of feedback control laws for active wing bending damping. Active wing bending damping reduces fatigue loads on the one hand, and reduces vertical accelerations in the cabin on the other. Additional feed-forward compensation of wing bending vibrations can provide further improvement of ride comfort and, assuming sufficient control authority, has the potential to also reduce peak loads. Best performance is obviously achieved by a combination of both active wing bending damping and feed-forward compensation of atmospheric turbulence excited wing bending vibrations. However, the performance of feed-forward control is very sensitive to plant uncertainties. Regarding wing bending vibration alleviation on large airliners, one major plant uncertainty is the fuel mass. A promising solution to overcome problems related to the sensitivity of performance in regards to plant uncertainties, is the adaptation of the feed-forward controller towards optimum performance using a least mean square algorithm. Such an adaptive feed-forward vibration control system has been successfully flight tested on the DLR ATTAS (Advanced Technologies Testing Aircraft System.) A reference for atmospheric turbulence was obtained by a nose boom mounted flight log sensor. Symmetrically commanded high bandwidth Direct Lift Control (DLC) flaps served as actuators. The obtained flight test data shows robust convergence of the adaptive feed-forward controller towards its optimum, as predicted by previous stability analysis of the adaptation algorithm. The ATTAS is a relatively small and stiff aircraft with a first wing bending frequency of 5 Hz. The converged feed-forward controller thus reduces the power spectral density of modal wing bending accelerations by only 20%, at least showing that the use of a conditioned alpha probe signal for compensation of atmospheric turbulence excited wing bending vibrations is feasible. A 20% reduction in terms of the power spectral density is not much, but the result is well in accordance with performance estimates drawn from rough calculations with the two-dimensional von Karman turbulence model. The same model predicts a 75% reduction of the power spectral density of modal wing bending accelerations (50% reduction of the magnitude of modal wing bending accelerations respectively) for large airliners where the first wing bending frequency lies at about 1 Hz. A 50% reduction of the magnitude of modal wing bending accelerations is comparable to what is achievable with active wing bending damping control systems today, thus showing that for large airliners the proposed feed-forward compensation of atmospheric turbulence excited wing bending vibrations provides a powerful method for further reduction of dynamic loads and improvement of ride comfort.