| With the continuous improvement of linear motion control accuracy in various fields,permanent magnet linear synchronous motor（PMLSM）,as the core component of linear motion system,has been more and more widely used in precision numerical control machining,semiconductor processing,rail transit,precision instruments and other fields owning to its superiorities of high speed,high precision,high power density and simple structure.But at the same time,due to the lack of buffering of the intermediate transmission mechanism,the PMLSM is greatly affected by the internal and external perturbations in the system,which in turn affects its control accuracy.Therefore,in order to enhance the robust performance of the system and improve the position tracking accuracy,a recursive nonsingular terminal sliding mode control（RNTSMC）method combined with neural network-based perturbation estimator,which has finite-time convergence characteristics is proposed in this thesis.Firstly,the composition and working principle of PMLSM are expounded,and the mathematical model of PMLSM with uncertain factors in dq coordinate system is introduced and established in detail.Then,the PMLSM vector control system is established based on the model,and the causes of perturbation in the system and its impact on the system are analyzed.Next,aiming at the problem that the traditional sliding mode control has large chattering and can not converge in finite time,a PMLSM nonsingular terminal sliding mode controller is designed,and a power-type reaching law with the advantages of exponential reaching law and power reaching law is also designed to ensure rapid convergence and suppress chattering.However,the simulation results show that the tracking accuracy of the nonsingular terminal sliding mode controller is poor when subjected to large disturbances,so sliding surface of the controller is used as the inner sliding surface,and on this basis,the outer nonsingular terminal sliding surface in integral form is designed.Both sliding surfaces consist of RNTSMC with double-layer sliding surface structure,and the outer and inner sliding surface converges in turn to ensure that the tracking error can converge to zero in the theoretical finite time under the condition of perturbation.The simulation results show that RNTSMC can effectively improve the mover tracking accuracy and make the system have better control performance.Finally,due to there are many perturbations in practical applications,which will affect the tracking accuracy of the system.In order to eliminate the steady-state error of RNTSMC in practical applications as much as possible,a double-hidden-layer radial basis function neural network and a continuous radial basis emotional neural network are designed as perturbation estimator respectively to estimate and compensate for the effects of lumped perturbations.The approximation effect of neural network is further improved by adaptively updating the connection weights,and the stability of the two strategies is verified by Lyapunov stability criterion.The simulation results show that both of the two proposed methods can improve the position tracking accuracy of the system.In contrast,the system that adds the continuous radial basis emotional neural network to compensate the perturbation has the better control performance. |
