基于永磁直线同步电机的光刻机掩模台鲁棒自适应神经网络控制  

A Robust Adaptive Neural Network Control Method Based on Permanent Magnetic Linear Synchronous Motor for the Reticle Stage of Lithography

作　　者：王一光[1] 李晓杰[2] 陈兴林[2] 

机构地区：[1]黑龙江大学,哈尔滨150080 [2]哈尔滨工业大学,哈尔滨150001

出　　处：《电工技术学报》2016年第6期38-46,共9页Transactions of China Electrotechnical Society

基　　金：国家重点基础研究发展计划(973计划)(10007.07-LB7);国家科技重大专项(2009ZX02207)资助项目

摘　　要：针对光刻机掩模台宏动永磁直线同步电机(PMLSM)提出一种鲁棒自适应神经网络轨迹跟踪补偿控制策略。该策略的主要特点是采用径向基函数(RBF)神经网络实时在线进行模型参数不确定性和各种外界非线性扰动的估计。为了验证策略的效果,建立基于参数不确定性和外界非线性扰动的掩模台宏动PMLSM模型,并针对此模型进行控制策略的规划、推导和稳定性理论分析,分析结果说明该策略可以保证位置和速度跟踪误差的收敛性。通过在光刻机掩模台上进行的五阶S曲线跟踪控制实验验证了此控制策略的实际控制效果,实验结果显示跟踪准确度达到了满意的效果。由于此控制策略的最大优点是不需要对实际物理系统参数和难于测量的外界不确定性扰动做精准的建模,所以非常适合应用在超精密运动控制领域。This paper presents a robust adaptive neural network tracking compensation control strategy based on permanent magnetic linear synchronous motor （PMLSM） for long-stroke reticle stage of lithography, It can estimate the model uncertainty and external nonlinear disturbance real-time online by radial basis function （RBF） neural network. A long-stroke PMLSM model of reticle stage based on parametric uncertainty and external disturbance was established. The derivation of the control strategy and the theoretical stability were analyzed. It was shown that the proposed model can guarantee convergence of the position error and velocity error. The actual effectiveness of this control strategy was verified by a fifth-order S-curve tracking experiment on the long-stroke reticle stage of lithography. The experimental data showed that the tracking accuracy met the design requirements well. This strategy doesn＇t require precise modeling of the actual system parameters and the external disturbances which are difficult to measure. It is very suitable for the application in precision motion control field.

关 键 词：永磁直线同步电机 光刻机掩模台 鲁棒自适应 神经网络 运动控制 

分 类 号：TM341[电气工程—电机]