永磁直线电机自构式模糊神经网络控制器设计

PMLSM self-constructing fuzzy neural network controller design

针对永磁直线同步电动机伺服系统中负载扰动和负载质量变化等参数不确定性对伺服系统动态性能的影响,设计自构式反馈模糊神经网络控制器。该控制器在模糊控制的非线性辩识功能和神经网络的自学习功能的基础上,各层神经元的个数可以根据误差状况发生改变,从而在保留了神经网络自学习能力的同时,增强了神经网络的实时性,提高了伺服系统的动态性能。自构反馈机制的引进,增强了神经网络的适应性。仿真结果表明,基于自构式反馈模糊神经网络控制器的永磁直线电机伺服系统对于参数的变化、外部干扰等具有较强的抑制作用,系统具有较强的鲁棒性。

A self-constructing feedback fuzzy neural network controller （SCFFNNC） for permanent magnet linear synchronous motors （PMLSM） is presented in the paper aiming at the influence of load disturbance and system parameters uncertainty to the PMLSM servo system dynamic performance. The controller combined the non-linear identification of fuzzy control with self-learning ability of neural network. The number of neurons for each layer of neural network can change on-line with the variation of error. The controller can reserve the self-learning ability, improve the real-time ability for neural network and en- hance the dynamic performance of servo system. The self-constructing feedback can improve the adaptation of neural network. Simulation results show that the PMLSM servo system based on SCFFNNC can suppress the parameters variations and external disturbances and is featured by strong robustness.permanent magnet linear synchronous motors; self-constructing