Neural network-based robust integral error sign control for servo motor systems with enhanced disturbance rejection performance. (October 2022)

• Record Type: Journal Article
• Authors: Ding, Runze; Ding, Chenyang; Xu, Yunlang; Liu, Weike; Yang, Xiaofeng
• Is Part Of: ISA transactions. Volume 129(2022)Part A
• Page Start: 580
• Page End: 591
• Publication Date: 2022-10
• Subjects: Neural network control -- Adaptive control -- Robust control -- Servo system; Engineering instruments -- Periodicals; Engineering instruments; Periodicals; Electronic journals; 629.805
• DOI: 10.1016/j.isatra.2021.12.026
• Languages: English
• ISSNs: 0019-0578

Abstract

Abstract: Uncertain dynamics and unknown time-varying disturbances always exist in servo systems and deteriorate tracking accuracy significantly. To tackle the problem, this paper presents a novel adaptive robust control scheme based on neural networks and the robust integral of the sign of the error (RISE) method. In the proposed scheme, a new neural network compensator is developed, where a reference-driven neural network and an error-driven neural network are employed to compensate for uncertain system dynamics and unknown time-varying disturbances, respectively. And an RISE-based robust feedback controller is designed to suppress uncompensated dynamics. Asymptotic tracking control of the servo system with uncertain dynamics and unknown time-varying disturbances is guaranteed by using the Lyapunov theory. Comparative experiments and simulations with different reference signals and various types of external disturbances were conducted based on a linear motor-driven stage. Experimental and simulational results verify the superior tracking performance and powerful disturbance rejection ability of the proposed method. Highlights: An NNRISE control method with enhanced disturbance rejection performance is proposed. An NN compensator consisting of two different NNs is developed. A continuous update law is designed to train the NN compensator online. Experiment and simulation results reveal the validity of the proposed method.