本研究利用MatLab/Simulink所提供之Real-Time Windows Target即時控制系統,設計平面磨床多軸精密定位控制,經由理論推導數學模型及控制法則,再以實驗驗證後其結果良好,均可滿足需求。在砂輪磨耗研究方面,將對磨床在磨削時產生的砂輪磨耗作分析,找出自動補償之方式。而為省去量測砂輪磨耗之感測器,先利用測量儀器擷取大量之砂輪磨耗數據,再透過類神經網路之設計,推導出砂輪預估補償的模型。並經由實驗證明所推導之砂輪預估補償模型與實驗值頗為接近,在某種精度之使用要求情況下,可提供一個藉由軟體分析的方法,以達到砂輪磨耗自動補償之目標。
This research makes use of the Real-Time Windows Target that MatLab/Simulink provides, to design the positioning control for multi-axes of a surface grinding machine. Modeling and control is developed and verified by experiment. Analysis of the wear of the grinding wheel is carried out, which attempts to find a method for automatic compensation. For saving the expensive measure sensor, a neural network method by means of numerous experimental grinding data is used to develop a predictive compensation model. The accuracy of the compensation model is verified acceptable for the predictive compensation value is approximate to the experiment. Hence, the proposed compensation model is allowed to provide a software tool that calculates the compensation value automatically within an acceptable accuracy requirement.