Vibration appears in many technological fields, sometimes as part of the process itself, and sometimes as an undesired consequence of the system function. In all cases, the presence of these oscillations can result in structural fatigue issues or in reduction of the system performance, like it is the case of tracking problems in robotic arms or measuring devices, like big moving telescopes. Aeronautical structures are especially sensitive to vibration problems. Their light nature and the presence of periodic and non-periodic perturbations, which are caused at a large extent by the propellers, are responsible for the transmission of vibration power inside the cabin resulting in noise and comfort problems to add to the previous mentioned fatigue issues. Vibration damping mechanisms are used in aeronautic structures, mainly as passive elements for isolating the sources of oscillation like rubbers or springs, or by means of tuned mass dampers. However, passive approaches have performance limitations. Their passive character makes them unable to adapt themselves to changing conditions, and they are normally restricted to certain frequencies of operation. For these reasons, active solutions have gained interest during the last years. These solutions require the use of actuators capable of entering power in the system. Examples of this trend are: piezoelectric patches for high frequency noise cancellation in cabin or in rotor blades; hydraulic actuators for structural damping, among others. The solutions normally apply speed or acceleration feedback signals for modifying the effective mass or structural damping. The approach we are describing in the present paper investigates the improvements that can be obtained by the use of more advanced monitoring capabilities for identifying the perturbation affecting the structure. This solution tries to avoid reacting to the effects of the perturbation, but compensating them as soon as possible, therefore minimizing the undesired effects. This is done by combining model based observers with sensors and actuators in the structure. In combination to the use of estimators, learning algorithms can improve the performance of the damping system by predicting the perturbation to arrive and therefore actuating at the same instant it appears. This approach is intended to reduce the sampling restrictions from pure estimating solutions.
doi: 10.12783/SHM2015/17