Adaptive Fractional-Order Sliding Mode Controller with Neural Network Compensator for an Ultrasonic Motor

TL;DR

This paper introduces an adaptive fractional-order sliding mode controller with a neural network compensator for ultrasonic motors, enhancing tracking accuracy and reducing chattering through fractional calculus and optimal control techniques.

Contribution

It presents a novel adaptive fractional-order sliding mode controller combined with a neural network compensator, addressing practical implementation issues with the short memory principle.

Findings

Improved tracking accuracy demonstrated in experiments.

Reduced chattering compared to traditional SMC.

Effective compensation of residual errors by the neural network.

Abstract

Ultrasonic motors (USMs) are commonly used in aerospace, robotics, and medical devices, where fast and precise motion is needed. Remarkably, sliding mode controller (SMC) is an effective controller to achieve precision motion control of the USMs. To improve the tracking accuracy and lower the chattering in the SMC, the fractional-order calculus is introduced in the design of an adaptive SMC in this paper, namely, adaptive fractional-order SMC (AFOSMC), in which the bound of the uncertainty existing in the USMs is estimated by a designed adaptive law. Additionally, a short memory principle is employed to overcome the difficulty of implementing the fractional-order calculus on a practical system in real-time. Here, the short memory principle may increase the tracking errors because some information is lost during its operation. Thus, a compensator according to the framework of Bellman's optimal control theory is proposed so that the residual errors caused by the short memory principle can be attenuated. Lastly, experiments on a USM are conducted, which comparative results verify the performance of the designed controller.