標題: 整合LuGre摩擦力模型設計應用於3C產業之高速微動控制器
Design the High-speed Micro-motion Controller for the 3C Industry with the LuGre Friction Model
作者: 黃琮昇
Huang, Cong-Sheng
徐保羅
Hsu, Pau-Lo
電控工程研究所
關鍵字: 3C;微動控制;CNC;LuGre;適應性擾動觀測器;前置控制器;3C;micro-motion;CNC;LuGre;adaptive DOB;feed-forward controller
公開日期: 2012
摘要: 有別於傳統CNC製造業,現今操作於3C製造業的微動控制相對高頻,而且具有命令較小、頻繁地通過零速度的特性,導致運動過程中受到摩擦力的影響而降低加工精度,控制上變得更加困難。傳統穩態摩擦力補償是藉由定速時量測得摩擦力大小值建立穩態摩擦力模型,再依不同命令進行補償,但只適用於命令以低頻緩慢變化時。實驗結果指出當命令很小時,使用線性控制器下CNC的Z軸幾乎靜止不動。 本論文採用LuGre摩擦力模型來克服摩擦力現象,並提出一套LuGre模型的係數找尋方法,能幫助使用者快速找到正確的係數。本研究以DSP F28335為控制核心實現穩態摩擦力補償、動態摩擦力補償、適應性擾動觀測器與前置控制器等架構,並分別探討各種控制架構的功用與特性,最後成功實現於AC伺服馬達,速度迴路頻寬由394Hz提升至663.7Hz,位置迴路頻寬在加上前置控制器後,由9.5Hz提升至113.8Hz,而LuGre摩擦力補償也實現於DYNA 1007 CNC工具機上,進刀時,追蹤誤差的方均根值改善率為67.5%,追蹤誤差的峰值的改善率為72.0%。
Micro-motion control in the 3C industry is operated in the relatively high-frequency range compared with traditional CNC processing. The control commands are difficult to precisely follow mainly because the motion command is relatively small and it passes through zero velocity frequently to cause the problem of friction. Traditional static friction compensation, which is obtained from a constant speed operation, is suitable only when the motion command is slowly-varying in a low frequency range. Experimental results indicate that the axis is nearly stop by applying linear controllers if the motion command is small. In this Thesis, the LuGre model is adopted and a practical and systematic approach to obtain the coefficients for the LuGre model is proposed. The DSP microcontroller TMS320F28335 is adopted for the AC servo driver of permanent synchronous motor (PMSM) in this study. All the static friction compensator, dynamic friction compensator, adaptive disturbance observer (DOB) are integrated and realized in this servo driver. By applying the present controller with the LuGre model, the bandwidth of the velocity loop is improved from 394Hz to 663.7Hz, and the position loop is improved from 9.5 Hz to 113.8 Hz, separately. The LuGre model has also been realized on DYNA 1007 CNC machine and results also show that the tracking error is improved 67.5% in RMS and the tracking peak error improvement is 72.0%.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079912565
http://hdl.handle.net/11536/49266
Appears in Collections:Thesis