Friction-free approach for a servo system: Design of a robust sliding-mode control based on a recurrent fuzzy neural network

Abstract

Uncertainties such as disturbance and nonlinear friction always exist in physical motion control systems and degrade the precision of their tracking performance. In this research, a practical control strategy is integrated with a continuously differentiable friction model for a high-performance AC servo motor that performs with a continuous control input and is, hence, more properly suited for industrial applications. To further lessen the external disturbance and improve the tracking accuracy, a recurrent fuzzy neural network scheme was developed to approximate the system uncertainties, and its stability was guaranteed by the design of robust laws. The proposed controller was implemented on a high-performance digital signal processor (TMS320C6727). The results of the tracking simulations and experiments show that the proposed control scheme has a feasible and effective design for engineering servo system. In comparison with other algorithms, our approach has smaller values for both the peak and the root mean square of the tracking error while maintaining friction compensation, and it can potentially have a software implementation.