Line-of-sight stabilization of an optical instrument using magnetostnctlve actuators is described in this study. Various stabilization methods, i.e. gyroscopic, hydraulic, piezoelectric, electrodynamic and magnetostrictive methods, are compared and magnetostrictive stabilization is selected for its relatively large stroke length, low input voltage and wide frequency bandwidth. The system makes use of two magnetostrictive actuators, one at each end of the optical instrument, mounted between the moving base and instrument. Each actuator is equipped with cylindrical rods of Terfenol-D, a highly magnetostrictive material. Field coils are wound around the rods to produce a strain in the rods, thereby exciting angular motion of the instrument. Actuator stroke length is enhanced by means of a hingeless gain mechanism, rod prestressing and field biasing. Dynamic characteristics of the system are modelled to facilitate actuator, coil and control system design. A linear, single-degree-of-freedom actuator model, in state-space and transfer function forms, is derived and coupled to a distributed model of the optical instrument, using the Rayleigh-Ritz method. Transfer functions between actuator coil voltages and instrument angular acceleration are derived. Normal mode shapes, natural frequencies and damping factors are predicted. Design concepts for bias field, prestress, actuator gain and optical instrument support structure, are discussed and the most suitable concepts are selected. The required actuator gain, rod length and diameter, prestress spring stiffness, coil resistance and inductance are calculated. System components are designed in detail and safety of the design is checked. The actuators are characterized quasi-statically to determine the saturation strain, linear range of operation and DC bias field. The system is dynamically characterized to obtain transfer functions between the coil voltage and instrument angular acceleration. The test setups are described and limitations of the setups are discussed. Test results are processed and discussed. A comparison with the modelled results shows that the model is highly inaccurate. Reasons for inaccuracies are given and updating of the model is motivated. An updated model is obtained from the experimental results. The model is divided into electrical and mechanical subsystem models. The SDOF actuator models are replaced with 2DOF models (one for each actuator) and coupled to the instrument and base models, using substructure synthesis. The electrical and mechanical subsystem models are subsequently coupled. It is shown that the updated system model is considerably more accurate than the original model. A linear, suboptimal, disturbance feedforward plus output feedback controller, with output integral feedback, is designed, implemented and tested. An H2 optimal controller is designed and modified to improve robustness. The controller model is coupled to that of a suboptimal observer. An output integral feedback loop is added to further improve robustness. The controller is implemented in digital filter form. The test apparatus and procedure are described. Test results are processed and discussed. It is shown that the LOS stabilization system achieves 80% of the required isolation, over a frequency bandwidth of 0 Hz to 100 Hz. A summary of the work done, conclusions that can be drawn from the results, problems encountered and recommendations for future work, are given.
Thesis (PhD (Mechanical Engineering))--University of Pretoria, 2006.