H型龍門平台之命令整型疊代學習與交叉耦合PID類神經網路同步控制

Iterative learning command shaping and cross-coupled PIDNN synchronous controller for H-type gantry stage

吳昆璋(Kun-Zhang Wu)

指導教授 ： 陳金聖

國立臺北科技大學/機電學院/自動化科技研究所/碩士(2014年)

https://doi.org/10.6841/NTUT.2014.00391

摘要

高精度運動控制龍門平台應用在許多工業製造上，其中包含太陽能電池、平板、微電子產業、航空製造與檢測。然而，同步誤差與輪廓誤差將影響工件加工品質，當其誤差過大會導致工件加工跳機，因此，其龍門架構運動平台之同步與輪廓誤差控制在高速命令下對於高精高速製造與檢測變成很大的挑戰。
H型龍門平台的動態存在一些不確定性，其中包含參數變動，外部干擾與摩擦力，將交叉耦合誤差模型整合到PIDNN控制策略，保證雙軸馬達之追蹤誤差與同步誤差同時收斂至零。此外，疊代學習控制被應用於命令整型來改善其輪廓誤差。在PIDNN之網路結構連接權重疊代更新式是依據線上自我學習機制來獲得，而共平面軌跡是透過疊代學習對輪廓命令做補償，最後從實驗結果證實所提出方法是有效性。

關鍵字

H型龍門平台 ； 同步誤差與輪廓誤差 ； 交叉耦合PID類神經網路同步控制器 ； 疊代學習控制

並列摘要

High precision motion control of gantry stage has found numerous applications in the manufacturing industries including, solar cell, flat panel, microelectronics and aerospace manufacturing and inspection. However, the synchronous and contouring errors will influence the working quality of work-pieces and even lead to reject the work-pieces due to the over current protection. Hence, the control of the synchronous and contouring error in the motion stage of gantry configuration has become a challenge with increasing demands for high-speed and high accuracy manufacture and inspection.
The dynamics of the H-type gantry stage with a lumped uncertainty, which contains parameter variations, external disturbance and friction force, is introduced. Then, to guarantee both of the position tracking and synchronous errors of dual motors converge to zero simultaneously, hence, a cross-coupled error model is derived and incorporated into the proposed PIDNN control scheme. Furthermore, the iterative learning control is applied for command shaping to improve the contouring error. In PIDNN, the connective weights of network structure are obtained by on-line learning algorithms. In contrast, the compensated contouring commands are generated by iterative learning for specified coplanar trajectories. Finally, some experimental results are illustrated to depict the validity of the proposed control approaches.

並列關鍵字

H-type gantry stage ； synchronous and contouring error ； Proportional-Integral-Derivative Neural Network controller ； Iterative learning control

參考文獻

16.Y. Zhang, P. Sen and G. E. Hearn, “An on-line trained adaptive neural controller,” IEEE Control Systems, 1995, pp.67-75.

1.黃聖原，雙伺服CNC工具機新主從控制方法之研究與應用，碩士論文，中原大學機械工程學系，2006。

3.楊宗翰，龍門式雙軸平台之模糊類神經網路同步控制，碩士論文，國立台北科技大學自動化所，2011。

4.楊君賢，具機構耦合之雙線性伺服系統鑑別與控制，碩士論文，國立成功大學機械工程學系，2003。

5.S. Kim, B. Chu, D. Hong, H. K. Park, J. M. Park and T. Y. Cho, “Synchronizing dual-drive gantry of chip mounter with LQR approach,” IEEE International Conference on Advanced Intelligent Mechatronics, 2003, pp. 838-843.