一般而言，CNC 工具機以達到高速高精度為目標，影響加工精度方面原因很多，其中熱變位的影響約佔了整體誤差的 50% 以上。因此，預測熱變位以及進行補償就成為工具機達到高精度的重要因素之一。本文首先量測在不同主軸轉速工況下，各鑄件的溫度變化趨勢以及刀尖點與平台之間相對位移量，接著建立一套溫度計的排序與篩選機制，找出關鍵的溫度量測點，最後透過類神經網路建立機台主軸的熱變位補償模型，用於驗證本文建立之溫度計排序與篩選機制，並能提高熱變位補償模型的預測精度。

In general, CNC machine tools aim at achieving high speed and high precision. There are many factors in affecting machining accuracy and the influence of thermal displacement accounts for more than 50% of the overall errors. Therefore, predicting thermal displacement with compensation algorithm becomes one of the important factors for high precision of machine tools. This paper first measures the temperature changes of each casting components and the relative displacement between the tool tip point and the platform under different spindle speed conditions. Then a sorting and screening mechanism of the thermometer is established to determine the key temperature measurement points. Finally, the thermal displacement compensation model of the machine spindle is developed by using the neural network. It is used to verify the thermometer sequencing and screening mechanism established in this paper, and it is proved that the mechanism can improve the prediction accuracy of the thermal displacement compensation model.