Robust adaptive dynamic surface control using recurrent cerebellar model articulation controller-based function link neural network for two-axis motion control systems

This paper proposes a robust adaptive dynamic surface control system (RADSCS) using recurrent cerebellar model articulation controller-based function link neural network (RCMACFLNN) for identification and control of uncertain two-axis motion control system driven by two permanent-magnet synchronous motors (PMSMs) servo drives. The proposed control scheme incorporates a dynamic surface controller (DSC), a RCMACFLNN uncertainty observer, a robust controller and an optimal controller. First, an optimal computed torque controller (OCTC) is deigned to stabilize the two-axis motion control system. However, the control performance may be destroyed due to parameter uncertainties exist in the OCTC law for the reason that the linear optimal control has an inherent robustness against a certain range of model uncertainties. Therefore, to improve the robustness of the two-axis motion control system, an RADSCS is designed to achieve this purpose. In the RADSCS, the DSC is used to overcome the explosion of the complexity in the backstepping design to enhance the robustness of the two-axis motion control system. The RCMACFLNN uncertainty observer is designed to adaptively estimate the nonlinear lumped parameter uncertainty terms, yielding a controller that can tolerate a wider range of uncertainties whereas the robust controller is designed to recover the residual of the approximation error of the RCMACFLNN. In addition, the optimal controller is used to minimize a quadratic performance index. The online adaptive control laws are derived using the Lyapunov stability analysis and the optimal control technique. From the experimental results, the motions at X-axis and Y-axis are controlled separately, and the dynamic behaviors of the proposed RADSCS with RCMACFLNN can achieve robust and optimal tracking performance against parameter uncertainties.