磁浮平面电机精密LVRC运动控制理论与技术研究

Maglev planar motor, is the core device of the EUV lithography equipment for high-grade IC chip manufacturing due to its integrative driven, multi-DOF motion even in the vacuum environment. With strong dynamic coupling and obvious flexibility, maglev planar motor is influenced by various disturbances such as trajectory change, force ripple and magnetic field distortion. Therefore, it is usually difficult to design motion controller for maglev planar motor to meet the strict motion requirement of EUV lithography tools. In this proposal, research will be conducted to address the motion control problem of maglev planar motor. Firstly, the strongly coupled dynamics and the flexibility characteristic are analyzed to build MIMO model and realize dynamics decoupling, and a model based robust control strategy will be designed to guarantee system stability of the system. Secondly, since traditional linear control cannot guarantee excellent tracking error in low/high frequency at the same time, the variable gain control strategy is employed to optimize the error in multi-band frequency according to its frequency characteristics. Thirdly, to further suppress the influence of various disturbances, the learning trajectory compensation strategy is introduced to improve both the system response rapidity and the tracking accuracy. Based on the above mentioned control procedures, a Learning Variable Robust (VRC) control framework is synthesized to guarantee excellent motion performance for the maglev planar motor in the fields of “stability”, “accuracy” and “rapidity”. This project will provide theoretical innovation and key technology support for the industrial application of maglev planar motor.

磁浮平面电机具有一体化驱动、多自由度运动、适合真空环境等特点，是高端IC芯片制造装备EUV光刻机的核心部件。磁浮平面电机动力学强耦合、柔性特征明显，且受变轨迹、推力波动、磁场畸变等多源干扰因素影响，通常难以对其进行运动控制以满足EUV光刻的苛刻运动性能需求。本项目围绕磁浮平面电机高性能运动控制难题展开研究，首先考虑其多自由度强耦合动力学以及柔性特征，建立电机MIMO模型并实现动力学解耦，设计基于模型的鲁棒控制环节保证系统稳定性；鉴于传统线性控制无法兼顾高/低频性能，提出变增益控制策略，根据误差频率特征进行多频段误差综合优化；为进一步抑制多源干扰的影响，设计学习型轨迹补偿方案以提升电机响应快速性及轨迹跟踪精度。最终将上述策略统一于学习变增益鲁棒控制(LVRC)控制框架中，从“稳”、“准”、“快”三个方面综合保证磁浮平面电机的运动性能，为磁浮平面电机的产业化应用提供原理创新及关键技术支持。

磁浮平面电机是一种新兴的精密运动机构，具有无润滑摩擦、单一动子实现多自由度运动、可在真空环境运行等特点，是高端IC芯片制造装备EUV光刻机的核心部件。磁浮平面电机电磁驱动机理复杂，各轴运动具有强耦合性，且在实际工况中需考虑模型/轨迹变化和外在扰动等多方面因素影响，因此现有控制方法难以满足EUV光刻机的苛刻运动性能需求。本项目针对磁浮平面电机的精密运动控制难题开展系列研究。在电磁驱动机理方面，提出了一种二维周期磁场下无铁芯永磁平面电机的统一推力模型，并进一步考虑边缘磁场提出了扩展统一推力模型，能够描述任意形状和任意姿态的线圈在任意磁场谐波下产生的推力，补充完善了磁浮平面电机的电磁驱动建模理论体系。在超精密运动控制方面，提出了基于GRU(Gated Recurrent Unit)的轨迹修正量准确预测及前馈补偿控制、LARC(Learning Adaptive Recurrent Control)智能学习自适应鲁棒控制等策略，实现了和ILC(Iterative Learning Control)相近的控制精度，避免了对不同轨迹的迭代过程，不受限于重复轨迹，同时具有对外在扰动和模型变化的鲁棒性。在多轴轮廓运动控制方面，提出了一种基于数值方法的轮廓误差估计及补偿方案，建立了多轴轮廓误差解析模型，进一步提出了一种基于预测模型的在线迭代轮廓误差补偿控制方法，在轮廓运动控制中实现了和ILC相近的运动精度以及对轨迹变化的泛化能力。基于该项目，发表有本项目基金标注的SCI论文23篇(含SCI源)、EI论文4篇，申请发明专利6项(已授权2项)，为磁浮平面电机的进一步应用提供了理论依据和技术支撑。