適用於CNC工具機之參考命令型疊代學習控制器與背隙補償

Abstract

在工業4.0的時代背景下，智慧機械的概念不斷被提倡，在無人化工廠及產線全自動化背後，對於生產機械之精密度有更高的要求。現今常見的控制架構中，疊代學習控制器是一個架構簡單、效果顯著的控制方法，其可應用在具有重複性命令的工作中，藉由重複執行同一命令來提升機台的控制精度，目前被廣泛應用在CNC工具機以及機械手臂上。本研究將疊代學習控制器應用在已具有控制器的CNC工具機上，由於市售之控制器通常為封閉系統，使用者無法變更其控制架構，因此本研究以外掛疊代學習控制器的形式將其放置在控制閉迴路外，透過修改其參考軌跡命令來提升其控制精度。一般CNC工具機常用的控制命令為G code，而疊代學習控制器之控制輸入為時域位置命令，故本研究提出一套將G code轉換成時域命令的方法，並加入滾珠螺桿之背隙補償來改善其軌跡輪廓誤差。最後將其實現在東台搭載Fanuc控制器之CNC車床iTC-2000上，比較其與原始Fanuc控制器之追跡誤差及循圓測試的改善率，其測試結果在追跡誤差有顯著的改善，而循圓測試部分，真圓度也有8.3%的改善率。

Nowadays, industry 4.0 is the most popular issue of industrial manufacturing. As the basis of intelligent manufacturing, precision control of machine tools becomes more important than before, especially for unmanned and fully automated factories. Iterative learning control(ILC) is a commonly used control algorithm on robot arms and CNC machines. For those machines that perform the same task repeatedly and under the same operating conditions, ILC presents good performance with a simple architecture. In this research, we improve the performance of a CNC machine with a Fanuc controller on it by the ILC algorithm. Since the Fanuc controller is a closed system, users cannot change the control input to the CNC machine. Therefore, we put the ILC algorithm outside the control loop to modify the G-code commands to the Fanuc controller. To do so, we first transform the G code to the time domain position command. Second, we modify the command by the command-based ILC algorithm. Finally, we add backlash compensation to improve the contour error. We do experiments on Tongtai CNC machine iTC-2000 and the control performance is compared with that of the Fanuc controller. Experimental results show that ILC can decrease tracking error significantly, and the roundness of the circular test is improved by 8.3%.