線性馬達運動系統在巨觀及微觀區下之摩擦力分析及補償

Abstract

本文研究如何精密控制一個受摩擦非線性影響下的直線馬達驅動運動系統。 研究中發現 LuGre 模型，在描述微觀區摩擦力行為存在有若干的缺失，因此針對此區，提出了一個改善的磨擦力模型，以便更精準地預測其行為。 並利用此改善後的模型設計一前饋式摩擦力補償器，實驗顯示此摩擦力補償器在微觀區下明顯優於以LueGre 模型設計的補償器，且在巨觀區時，兩者具有相同的性能。 另外在巨觀區的實驗當中，我們也加入干擾觀測器消除一些系統的不確定性和外在的干擾，實驗也顯示系統的性能獲得大大地提升，可是卻發現干擾觀測器在作定位控制時，會產生極限循環而嚴重影響定位精度，因此本論文提供了一個避免干擾觀測器產生極限循環的充分條件，並據此條件設計控制器參數。 可是這些參數往往會降低系統的性能而減少實用性，故論文裡也提出了一個調適的機制既可以維持系統的性能也可避免極限循環的發生。

This paper studies how to control a linear-motor-driven motion system under the effect of nonlinear friction. The LuGre model has drawbacks and can’t describe some friction phenomena under micro-dynamic scale. Therefore, a modified friction model which can actually describe those phenomena is proposed. Then, a feedforward type friction estimator designed from this model is used to compensate the friction, and the experimental results show the compensator based on the modified friction model is much superior to that based on the LueGre model in micro scale region but have the same performance in macro scale region. In addition, we also add disturbance observer into servo loop to cancel the effects of system uncertainty and other disturbance under macro-dynamic scale. The experimental results show disturbance observer can certainly improve the system performance. Nevertheless, disturbance observer generates limit cycles in regulating control, which serverly decreases positioning accuracy. Therefore, this thesis derives a sufficient condition for the absence of limit cycles and exploits the condition to design controller parameters. The parameters designed from the condition usually make the system performance very low, so this thesis also introduces an adaptive mechanism that not only maintains the system performance but also eliminates limit cycles.