This thesis presents the repetitive control (RC) design of a novel linear magnetostrictive actuator. A repetitive controller is developed and tested on a novel linear magnetostrictive actuator to improve the tracking accuracy of the actuator to a periodic signal. The repetitive controller is designed based on an estimated model of the system which does not consider the model nonlinearities. In the repetitive controller, a learning controller helps the actuator track the reference signal faithfully. A Butterworth low-pass filter is designed to stabilize the system in high frequencies. The parameters of the learning controller and the low-pass filter are initially tuned during simulation and then tested in the experiment. The stability, robustness, convergence rate, and performance are discussed in the RC design, The trade-off between robustness and performance is taken into serious consideration. In the experiment part, sinusoidal and triangular reference signals are used to test the tracking performance of the repetitive controller. Simulation results show that the maximum tracking error to a sinusoidal signal with a magnitude of 1 mm could be limited to 0.1 um, which is 0.01% of the reference input. Experiment results demonstrate that the maximum tracking error to a sinusoidal reference signal with a magnitude of 0.5 mm and frequency of 0.1 rad/s could decrease to 20 um, which is 4% of the reference signal. Compared with the previous result, the maximum tracking error to a sinusoidal reference signal with a magnitude of 0.5 mm decreased from 100 um to 20 um. The experimental results demonstrate that the repetitive controller works effectively to improve the tracking accuracy of the linear magnetostrictive actuator.
