Position tracking performance enhancement of linear ultrasonic motor using learning control techniques

Iterative learning control (ILC) is known to be effective in improving the performance of tracking periodic trajectories. This control technique gives satisfactory tracking performance provided the task is strictly repeatable. Many a times, it is required to track periodic trajectories of different frequencies that are identical in spatial patterns. In such applications, any change in the reference trajectory requires initiating a fresh learning process for the conventional ILC, which results in larger error during learning phase and longer error convergence time. An improvement in tracking performance in such cases can be realized using the previously stored history of control efforts that gives optimum tracking performance. This knowledge helps in predicting the control effort for a different frequency but identical spatial pattern reference trajectory. In this paper, such a direct learning control (DLC) technique that uses the knowledge of the control history is proposed and subsequently used in co-ordination with the ILC for achieving excellent tracking performance and fast error convergence. Experimental results for position tracking of linear ultrasonic motor (LUSM) demonstrate that this hybrid scheme limits the tracking error and error convergence time to about one third as compared to that without using DLC.