Closed-Loop Subspace Identification for Self-Tuning Vibration Control of Structure Integrated Antenna Arrays

Kurzfassung

Technologies for integration of antenna arrays for radar, GPS and communication purposes into composite structures are a current topic of research. Especially in military aircraft applications a conformal antenna design is desired to save mass and installation space. Nevertheless, distortions of single antenna orientations caused by vibrations of the supporting structure lead to a decreased performance of the entire antenna array due to unknown phase differences. Smart-structures are the key technology for the reduction of undesired vibrations. By integration of sensors and actua-tors into the structure, a controller is capable of calculating control signals, based on sensor mea-surement, which eliminates the disturbing vibrations. Within the scope of NATO research task group SET-131 "Vibration Control and Structure Integrati-on of Antennas" a common demonstrator is planned. The aim of cooperation between Fraunhofer FHR, DLR and EADS within the scope of SET-131 is to build up a demonstrator consisting of a conformal antenna array with active vibration compensation using piezo actuators. The demonstra-tor antenna front-end will comprise seven circularly polarized antenna elements in the GPS fre-quency range for a Controlled Reception Pattern Antenna (CRPA) navigation application. A curved glass and carbon fiber compound will be manufactured to be integrated into the structure of an aircraft fuselage. For control of structural vibrations a number of piezo actuators and accelerome-ters will be integrated at suitable locations. For synthesis of high performance controllers, the model of the controlled plant, from actuator in-puts to sensor outputs, has to be as accurate as possible. Especially in composite shell structures the modal properties like eigenfrequencies and damping are very sensitive to temperature chan-ges. Therefore, feedback controllers are usually designed for specific operating points. The model of the controlled plant at each operating point is determined by off-line system identification of the undisturbed structure. In military aircrafts the fuselage temperature may vary over a wide range, hence the number of operating points has to be large for an acceptable coverage of the entire ran-ge. This paper presents a self-tuning H-infinity controller approach based on in-flight closed-loop sys-tem identification of the controlled plant to overcome the operating point formulation and to genera-te accurate and actual models. Starting with a nominal model, the plant is updated in specific time intervals. The controller is afterwards updated with automatic controller synthesis based upon the actual plant. The challenge of in-flight system identification is the permanent presence of stochastic disturbance of the structure. Sources of the disturbance are the turbulent boundary layer (TBL) and the engines. For system identification additional band-limited uncorrelated noise signals are added to the control signals. During this multi-reference test the plant in- and outputs are recorded. By means of a two stage sequential LQ-decomposition and subspace methods the deterministic part of the system is identified. The approach is experimentally tested at an active plate demonstrator. Identification is carried out with different number of samples, sizes of block Hankel matrices and ratios between identification and disturbance signals. Results show that the presented approach is applicable for in-flight controller tuning.