Application of Koopman-Based Control in Ultrahigh-Precision Positioning
Application of Koopman-Based Control in Ultrahigh-Precision Positioning
复制标题
基于库夫曼的控制在超高精度定位中的应用
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
I. Mezić
中科院分区：
文献类型：
--
作者：
S. Zelenika;E. Kamenar;Milan Korda;I. Mezić
超高精度定位装置在微系统技术和精密工程中具有极其重要的意义。相对运动中的机械元件的摩擦干扰往往限制了其定位性能。如果以纳米定位精度和准确度为目标，则必须通过适当的控制算法来识别、建模和补偿摩擦干扰。合适的实验装置，因此，研究摩擦扰动的影响。与国家的最先进的摩擦模型的参数进行了实验确定。不同的控制算法，如PID控制器，前馈控制器和自适应控制器，实验和数值验证和比较。事实证明，自适应控制器可以实现纳米级精度的定位，但在点对点定位应用中，可能会产生较大的过冲和与较长的建立时间相关的问题。结果表明，这些问题可以通过采用Koopman为基础的模型预测控制，允许简化建模负担，同时成功地补偿摩擦的影响，最大限度地减少。
Ultrahigh-precision positioning devices are of outmost importance in microsystems’ technologies and precision engineering. The frictional disturbances of mechanical elements in relative motion often limit their positioning performances. If nanometric positioning precision and accuracy are aimed for, frictional disturbances have thus to be identified, modeled and compensated for via appropriate control algorithms. Suitable experimental setups are therefore employed to study the effects of frictional disturbances. The parameters related to state-of-the-art friction models are experimentally identified. Different control algorithms, such as a PID controller, a feedforward controller, and adaptive controllers, are experimentally and numerically validated and compared. It is proven that adaptive controllers enable nanometric precision positioning, but in point-to-point positioning applications can give rise to large overshoots and issues related to lengthy settling times. It is shown that these problems can be minimized by employing the Koopman-based model predictive control that allows simplifying the modeling burden while successfully compensating the frictional effects.