ABSTRACFAn experimental implementation of a new distributed parameter shape control design methodology is presented. The technique is based upon the input/output representation of distributed parameter systems in a spatially-and temporallytransformed frequency space. The analysis is specialized to shape control through the introduction of generalized spatial transforms of the plant response which explicitly parameterize the shape control task. Experiments are summarized wherein a 4Oin pinned-pinned steel beams shape is controlled, in the presence of quasi-static and resonant disturbances, over a bandlimited set of four sinusoidal shape basis functions at a closed-loop temporal bandwidth of 2Hz, using distributed piezoelectric actuators. Temporal compensation is provided by digitally-implemented LQG/LTR compensators. A novel inner-loop damping formulation, based upon the second method of Lyapunov, is developed and implemented to damp the beam resonant response beyond the LQG/LTR control bandwidth.
. INTRODUCFIONA principal issue in the design and operation of deformable mirrors or large space-based large antenna reflectors is the precise control of the system's shape to guarantee satisfactory While the need to control the structural shape is clear, the practical implementation of shape control strategies is a subject of current research6'7'8. The link between the loss of performance and deformation of the structure hs typically been based upon the RMS value of its surface distortion. In this vein, Weeks9 has presented an analytically accessible study of structural shape control, using an integral equation formulation. The analysis is restricted to the static, open-loop problem, using discrete point displacement sensors and force actuators. Spatial perfonnance is posed in terms of determining actuator sites to best achieve (in an RMS sense) a single shape. Schaechter10'11'12 has implemented an experiment utilizing Weeks' analysis to drive a flexible beam, suspended vertically in a gravity field, to two static shapes. The experiment was static, and quasi open-loop. Certainly, if a structural control system can achieve some specific shape, then it will be capable of rejecting this same shape. However, the performance of any open-loop approach will suffer from plant model errors, as well as the quasi-static and dynamic nature of the disturbance environment.Although a reduction in the RMS surface profile error has a significant effect on the system perfonnance, it does not necessarily improve performance parameters specific to the details of the distortion throughout the surface of the reflector.These issues have recently been addressed for space structural systems in a static, open-loop optimization approach5. A more general dynamic approach to structural shape control has recently been developed by Burke13'14. In this approach, which is a generalization of previous optical investigations15'16, a structure's surface profile y(x,t) is parameterized by an expansion in an orthogonal set of "eigenshapes" ((x)) defined ove...