基于对角递归网络补偿器的直线永磁同步电机伺服系统

Servo System for PMLSM Based on Dynamic Recurrent Neuro Network Compensator

针对直线永磁同步电机伺服系统负载阻力扰动、动子质量变化等扰动提出了一种基于动态对角递归网络补偿器的位置伺服控制方案。利用动态对角递归网络具有内部反馈结构和动态映射能力对系统参数变化和扰动具有较强鲁棒性的特点,设计出一种新型非线性补偿器,并利用一种改进的递推预报误差学习算法加快权值调整速度、节约在线训练时间。

For conquering the disturbance of loading resistance,mass variation of active-cell and so on,a position servo control strategy based on dynamic recurrent neuro network compensator for PMLSM is proposed.A nonlinear compensator is designed by dynamic recurrent neuro network,which has relatively robustness on parameter variations and load disturbances on aid of interior feedback structure and dynamic mapping ability.Meanwhile,a mending RPE algorithm is proposed to shorten training time by improving the adjusting velocity of weights.