数控机床直线同步电动机磁悬浮系统的神经网络直接自适应控制

Direct Adaptive Control of Neural Network of Magnetic Levitation System of CNC Machine Tool Linear Synchronous Motor

针对数控机床可控励磁直线同步电动机磁悬浮系统的强非线性、外部扰动不确定性的问题,设计基于RBF神经网络直接自适应控制器。通过分析磁悬浮系统的运行机理,推导运动方程及悬浮力方程,进而建立系统的状态方程;用悬浮高度的跟踪误差和误差的变化量构造误差函数,设计直接自适应理想控制器并采用RBF神经网络对其进行逼近;设计自适应律来估计神经网络理想权值,对误差函数的变化率构造二次型Lyapunov函数,利用Lyapunov稳定性理论来证明系统稳定;通过Matlab对控制系统进行计算机仿真,结果表明该方法设计的控制器与自适应模糊滑模控制器和PID控制器相比,空载启动时调节时间减少了23.5%,突加负载时动态降落减少了64.7%,恢复时间减少了38.2%,具有稳态误差小,调节时间和恢复时间短,抗扰性较强的优点,能有效提高磁悬浮系统的控制性能。

Aiming at the problems of strong nonlinear and uncertainty of external disturbance of the controllable excitation linear synchronous motor magnetic levitation system of CNC machine tools,a direct adaptive controller based on RBF neural network is designed.By analyzing the operating mechanism of the magnetic levitation system,the motion equation and levitation force equation are derived,and then the state equation of the system is established.The error function is constructed by the tracking error of the levitation height and the variation of the error.The direct adaptive ideal controller is designed and approximated by RBF neural network;An adaptive law is designed to estimate the ideal weight of neural network.The quadratic Lyapunov function is constructed for the change rate of error function,and the stability of the system is proved by Lyapunov stability theory.The control system is simulated by Matlab,and the results show that compared with the adaptive fuzzy sliding mode controller and PID controller,the adjusting time of no-load starting is reduced by 23.5%,when the load is suddenly applied,the dynamic descent is reduced by 64.7%and the recovery time is reduced by 38.2%,it has the advantages of small steady-state error,short adjustment time and recovery time,strong disturbance immunity,and the control performance of the magnetic levitation system can be effectively improved.