Adaptive Hybrid Neural Network Sliding Mode Control of Ultrasonic Motors

Nonlinear disturbances such as viscous friction, static friction and external disturbances exist in ultrasonic motors, which seriously affect their high-performance motion control process. To realize the high precision scanning motion of ultrasonic motor, an adaptive hybrid neural network sliding mode control (AHNN-SMC) method is designed. The neural network basis function part adopts a combination of Gaussian basis and partial friction model, combined with an adaptive algorithm to achieve online nonlinear compensation of the ultrasonic motor. In order to solve the problem of overestimation of switching control gain caused by the difficulty of accurate estimation of uncertain upper bound in the control process, adaptive switching gain is further adopted in this paper to cut down the influence of chatter in the control process. In order to verify the performance of AHNN-SMC in ultrasonic motor scanning motion control, position tracking simulation experiments are performed for sinusoidal signals with a maximum displacement of 20mm and compared with several algorithms. The results show that the maximum tracking error (MAE) of AHNN-SMC is 3.15 μ m and the mean error (RMSE) is 1.25 μ m, and the results indicate that the method has good position tracking performance.