學習自動機在運動控制上的應用

Abstract

在精密製造工業中，切削過程的加工精密度和軸的定位扮演著相當重要的地位。然而，由於控制系統參數的變化和外界負載擾動（如切削力、摩擦力），影響到數控工具機的加工精密度。一般製造工業中，CNC工具機等許多的機器設備已經商品化，若要更改機器內部控制器的可能性不大。因此，我們可藉由對機器內部參數的調整，來提高加工精密度。本文將由控制系統能數對控制系統的影響，推導出一有效率的控制參數自動調整法則。 近年來，已經有許多人提出參數調整法則來改善控制系統的響應。但是，主要的缺點就是必須得到控制系統的數學模式。本文所提出的參數自動調整法則，將有效地解決這一個缺點。此法則以學習自動機(Learnig Automata)為基礎。最大的優點，在於不需要事先得知整個控制系統的數學模式，只需輸入和輸出的資料即可，而省下許多花在做系統判別與計算機計算的成本和時間，是一種既快速又實用的自動調整法則。在個人電腦上的模擬，乃實際應用在X-Y位置平台循圓軌跡測試，可發現其具有快速學習，並且有效地降低95%以上的輪廓誤差，證實了此參數自動調整法則的可行性。

In precise manufacturing industries, the accuracy of machining process is the most important requirement. However, in practice, the external disturbance (friction, backlash, ...etc.) or variation of the system parameters will decrease the accuracy of machining process. Therefore, it is crucial to significantly reduce the influence of external disturbance and properly set the system control parameters in the precise motion control. This thesis will introduce the method for tuning the system control parameters. In recent years, several approaches have been developed to tune the system control parameters for improving the motion performance. Among these methods, the major disadvantage is that the transfer function of the plant has to be available in advance. The major purpose of this thesis is therefore to present a simple tuning method without the knowledge of the plant model by applying the Learning Automata method. This method is capable of improving performance in the control system whose dynamics is unknown and only the input and output data of the control system are needed. This method will be applied to the DC servo control system. Simultion and experimental results will show the fast learning and the contouring error will be reduced over 95% after tuning. Results have proven that the proposed method is realizable in motion control.