反覆式學習控制應用於伺服馬達鈍齒效應之補償

Abstract

本論文利用反覆式學習控制（Iterative Learning Control）理論設計一伺服馬達控制器，以補償伺服馬達在轉動時所產生的鈍齒效應（cogging force）。利用德州儀器公司所生產的數位訊號處理器（DSP）TMS320VC33為系統的核心處理器，搭配μC/OSII即時多工作業系統，以Code Composer v3.0當作開發環境來透過運動控制卡進行命令產生以及加減速之規劃，實現反覆式學習控制器。 反覆式學習控制必須應用於具有反覆性或週期性的系統，文中將以運動時間與學習次數作為二正交向量來架構二維控制系統，使其能夠完整的描述控制系統本身的動態行為與學習過程的特性。將PC端所儲存之運動歷史資料透過反覆式學習控制器來更新下一次運動之控制輸入，在有限次數的實驗內達到良好的運動控制。

In this thesis, a servo motor controller is designed based on Iterative Learning Control（ILC） algorithm to compensate the cogging force generated by the rotation of the servo motor. By using the digital signal processor（TMS320VC33） produced by Texas Industrial as the kernel processor of the system, μC/OSII as the real-time multitasking operating system, and Code Composer v3.0 as the development environment, the command generation, acceleration and deceleration control through the motion card are realized, and the Iterative Learning Controller is implemented. The Iterative Learning Control must be used on a repeatable or a peridocital system. By using motion time and iteration times as the two orthogonal vectors, a two-dimensional control system is construct, which can discribe the attribute of the dinamic behavior and learning process completely. The next control input is updated by the Iterative Learning Controller through the history data of motion on PC, and a well motion control result is achieved in limited experiments.