精密機台水平監控系統

水平精度是影響加工精度的一個關鍵因素。如果機器的水平精度不佳，可能導致機床底座上的負載分布不均勻而導致機床底座因為重力產生變形。這種變形會影響設備運動的直線度和角度精度，最終對加工平整度產生負面影響。目前大部分的研究集中在用於離線的水平精度檢驗，對於實時監測機制的關注較少。為了應對這一問題，本研究提出了一種基於精密設備的實時水平精度監測系統，該系統包括具有力感應能力的智慧腳座和一個電腦應用程式。該系統用於實時觀察腳座上的荷重數據，並計算目標平面的水平狀態。同時開發了一個使用者介面，供使用者監測實時腳座荷重與水平狀態。接著將建構各腳座的標準負載，即水平狀態下各腳座的荷重數值。最後通過人工智慧模型建立腳座負載和目標平面水平精度之間的關係。介面將會顯示各腳座的即時負載與標準負載、計算得到的水平狀態供使用者快速瞭解機台水平狀態。透過此種間接式的水平精度量測，本研究可以解決目前沒有即時水平精度檢驗方法的問題。透過實驗驗證，以每1mm/m或0.5mm/m為區間對水平精度進行劃分，可分別達到97.6%、90.5%之準確率，而應用回歸方法則可獲得平均殘差為0.0155°之確切角度數值。

關鍵字

工具機；人工智慧；水平精度；即時監控；物聯網

並列摘要

Horizontal accuracy is a critical factor affecting machining precision. A poor horizontal accuracy can lead to uneven load distribution on the machine base, causing deformation due to gravity. This deformation affects the linearity and angular accuracy of the equipment's movement, ultimately negatively impacting machining flatness. Currently, most research focuses on offline horizontal accuracy inspections, with little attention given to real-time monitoring mechanisms. To address this issue, this study proposes a real-time horizontal accuracy monitoring system based on precision equipment. The system includes smart footings with force-sensing capabilities and a computer application. It is designed to observe the load data on the footings in real-time and calculate the horizontal state of the target plane. Additionally, a user interface has been developed to allow users to monitor real-time footing loads and horizontal states. Standard loads for each footing will be constructed, representing the load values of each footing under horizontal conditions. Finally, an artificial intelligence model will be used to establish the relationship between the footing loads and the horizontal accuracy of the target plane. The interface will display the real-time load and standard load for each footing and the calculated horizontal state, allowing users to quickly understand the machine's horizontal state. Through this indirect horizontal accuracy measurement method, this study addresses the current lack of real-time horizontal accuracy inspection methods. Experimental verification shows that dividing the horizontal accuracy into intervals of 1mm/m or 0.5mm/m achieves accuracy rates of 97.6% and 90.5%, respectively. Applying regression methods yields an exact angle value with an average residual error of 0.0155°.