Auslegung und Vermessung eines aktiven Paneels mit optimal platzierter Aktuatorik unter Berücksichtigung des Realverhaltens adaptiver Steuerungsalgorithmen

Abstract

Active structures for noise and vibration reduction are increasingly used in the low frequency range in the aircraft industry. These structures are combinations of lightweight materials and integrated transducers such as piezoelectric ceramics. The poor damping properties of lightweight materials in the low frequency range are significantly improved by controlling the transducers. The placement of the integrated transducers within the lightweight structure is of crucial importance for the performance of the active system. In this thesis a concept is investigated, which integrates the real behaviour of an adaptive feedforward control system in the placement of piezoelectric transducers. The performance of an adaptive feedforward controller is determined by the causality of the system and the system order which is based on the filter weights, the reference sensors, the error sensors and the secondary sources (actuators). To develop an efficient system with minimum order the number and the positions of the sensors can be optimized. The optimization of the sensor positions is carried out by an energy cost function which is based on the observability Gramian. The performance of an adaptive feedforward control system in steady state is calculated by an optimal causal control system. Furthermore, the placement of the transducers is optimized by a genetic algorithm. The cost function of the genetic optimization algorithm is based on the optimal causal control system, so that the genetic optimization calculates the best locations of transducers which should be used by an adaptive feedforward control system. By simulating an adaptive feedforward control system of a CFRP panel is shown that an adaptive feedforward control system in steady state has got the same performance as an optimal causal control system. Furthermore, the genetic optimization algorithm is calculated for a FE modelled steel plate. In the optimization process two different cost functions are applied, the first is based on the optimal causal control system and the second is based on a non-causal frequency domain optimal control system. The location of the piezoelectric transducers and the performance of the active systems are different. Finally, a steel plate equipped with sensors and actuators is measured by experiment. The plate is controlled by an adaptive FxLMS-Algorithm. The result equal the prediction of the simulation achieved by the optimal causal control system.