摘要
对直线永磁同步电机伺服系统提出了一种基于动态对角递归网络补偿器的IP位置控制方案。在对比干扰观测器前馈IP位置控制器的基础上,利用动态对角递归网络(DRNN)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DRNN非线性补偿器,并提出一种改进的RPE学习算法加快权值调整速度、节约在线训练时间。仿真表明该方案能明显改善位置跟踪精度并增强系统鲁棒性。
A IP position control strategy based on DRNN compensator for PMLSM was proposed. After contrasting to IP position controller based on disturbance observer, a nonlinear compensator was designed by DRNN, which has relatively robustness on parameter variations and load disturbances on aid of interior Feedback structure and dynamic mapping ability. A mending RPE algorithm was proposed to shorten training time by improving the adjusting velocity of weights. The simulation indicates that this strategy can improve tracing precision and enhance the system robustness.
出处
《机床与液压》
北大核心
2008年第8期248-251,共4页
Machine Tool & Hydraulics
基金
国家自然科学基金项目(60772005)
关键词
直线永磁同步电机
动态对角递归网络
RPE算法
Permanent magnet linear synchronous motor
Dynamical recurrent neural network
Recursive prediction error (RPE)