直線馬達運動平台之參數鑑別與微小路徑定位控制

Abstract

本文研究如何於直線馬達驅動運動系統作精密控制微小路徑位移。在控制馬達之前，應建立精準的數學模型，並鑑別正確模型參數，方可設計理想的控制器頻寬，以達精密控制之目的。研究中建立兩種系統模型，一種為巨觀下的數學模型，包括易受控之LuGre 動態摩擦力模型，重要的系統參數如慣量、阻尼係數及庫倫摩擦力之模型參數等均可以用簡單的實驗來鑑別；另一種為微觀下的數學模型，利用 PRBS 測試訊號來鑑別刺毛未斷離狀況下的線性化數學模型。利用建構的兩種模型設計微觀下跟巨觀下的控制器架構，在微米等級及奈米等級的微小位移實驗中，可發現微觀控制器的追蹤誤差及定位誤差比巨觀控制器佳，都能達到20nm (1BLU)控制精度。若下達1000µm大位移的路徑命令，利用建構之兩階段控制器切換架構，從巨觀控制器切換到微觀控制器，可加快定位時間，本研究最慢將使用0.394秒到達20nm (1BLU)定位精度，若使用Bumpless 的切換方式，最快使用0.280秒到達定位20nm (1BLU)定位精度，切換過程中亦無過補的現象。若在大位移定位後，再走小位移路徑追蹤，微觀控制器的控制效果依然表現良好，能達到20nm (1BLU)追蹤精度。

Linear-motor-driven motion systems are becoming popular in recent years due to its simple structure and absence of flexible coupling. In this dissertation, the parameters of a linear-motor-driven motion system, such as the inertia, viscous damping coefficient, and Coulomb friction, can be identified more quickly and accurately by using disturbance observer. If the motion table is operated in the presliding region, the bristle deflection of the system is almost equal to the movement of the motion table. A linearized micro-scale model of the motion system is also proposed in this study. The parameters of the linearized model are obtained from a PRBS testing signal experiment. A micro-scale controller is designed based on the linearized micro-scale model, and by contrast, based on macro-scale model, three macro-scale controllers are designed. Comparing to the tracking and positioning experiments of the micro-scale controller and three macro-scale controllers, we obtain that the tracking and positioning error are about 1 BLU, 20nm by using the micro-scale controller under nanometer scale trajectory. For speeding up the positioning performance under macro-scale positioning movement, a two-stage controller switching algorithm from a macro-scale controller to a micro-scale controller is proposed in this study. As the motion near the desired target, the controller is smoothly switched from the macro-scale controller to the micro-scale controller. The choice of the switching condition can be the trade-off of the maximum position error (overshoot) and the settling time. In our experiment results, it totally takes at most 0.394 second for the motion table to reach the target position of 1000µm with the positioning accuracy of one BLU, 20nm. If we use the bumpless switching algorithm to switch two controllers, it totally takes at least 0.280 second for the motion table to reach the target position of 1000µm with the positioning accuracy of one BLU, 20nm and no overshoot phenomenon during switching. After a large scale positioning movement, using the micro-scale controller under micro-scale trajectories still has the best tracking performance within one BLU, 20nm.