Robust Adaptive Neural-Network Super-Twisting Sliding-Mode Control for PMSM-Driven Linear Stage with Uncertain Nonlinear Dynamics

This paper proposes a robust adaptive supertwisting sliding-mode control (RASTSMC) scheme for PMSM-driven linear stage control system with uncertain nonlinear dynamics to achieve high precision-positioning. First, a supertwisting sliding-mode controller (STSMC) is designed to stabilize the linear stage. However, the control performance may be destroyed due to the unknown model uncertainties, friction and backlash nonlinearities of the ball-screw, the parameter variations of the PMSM servo drive and external disturbances. Therefore, to improve the robustness of the control system performance, a RASTSMC is proposed, which incorporates a STSMC, a function-link interval type-2 Petri fuzzy-neural-network (FLIT2PFNN) estimator and a robust controller. The STSMC is adopted to reduce the chattering phenomenon, the FLIT2PFNN estimator is developed for the approximation of the unknown nonlinear dynamics online and the robust controller is designed to recover the residual of the FLIT2PFNN approximation errors. Furthermore, the online adaptive laws are derived based on the Lyapunov approach, so that the stability and robustness of the overall control system are guaranteed. A real-time implementation is carried out via dSPACE1104 control to verify the efficacy of the proposed control approach. Furthermore, the experimental results of the proposed RASTSMC shows good dynamic performance regardless of unknown model uncertainties and external disturbances.