Position tracking performance enhancement of linear ultrasonic motor with direct learning control technique

The ultrasonic motors are finding increasing demands in high precision positioning applications in industry. These motors are based on friction drive mechanism. It is difficult to formulate exact mathematical model of the motor drive system due to complex nonlinearities involved with friction and inverse piezoelectric phenomena. These nonlinearities pose significant problem for precise position control of the motor. In this paper, first we compensate the nonlinearity due to deadzone and a linear PI controller is used as the position controller. The performance of such a controller is evaluated experimentally and it is observed to be comparable to those obtained using computationally intensive nonlinear control schemes based on neural networks and fuzzy logics. For repetitive position tracking applications, the tracking error can be further reduced by augmenting an iterative learning control (ILC) to the existing, feedback controller. It is observed that this plug-in controller helps to reduce the tracking error by a factor of ten. Such ILC scheme works only if the task is strictly repeatable in nature. For nonrepeating trajectory tracking tasks, a direct learning control (DLC) scheme is proposed. Based on stored historical knowledge of control efforts, the control effort is predicted for new time scale reference trajectory. Experimental results obtained verify good position tracking performance of the proposed scheme.