Neural Modeling and Control of Hysteretic Dynamics in Ultrasonic Motors

Ultrasonic motors operate on the inverse piezoelectric effect and the frictional driven force. Complex nonlinear characteristics such as hysteresis and friction nonlinearities are involved. Due to the inherent undifferentiable property and complex contact mechanism, to design a high performance controller is a real challenge. To combat this problem, a novel neural modeling and control strategy is presented. The motor velocity is adopted to allow for the nonlinear friction effect, and the multi-valued hysteresis of the motor is transformed into a one-to-one mapping with the introduction of a special hysteretic operator based on the expanded input space method. Based on the neural model, an adaptive neural controller is designed, where a so-called generalized gradient of the hysteretic operator is introduced to get the sensitivity information. Finally, the simulation results are presented for the validation of the proposed approach.