摘要 隨著工業的進展，製造業目前已處於時間與品質並重的時代，因此目前自動化製造設備一直朝向高速高精度的方向發展。而CNC工具機在循圓或做曲面加工時，為了減少其追蹤誤差，適當的摩擦力補償量是工具機高精度加工所必要的。摩擦包含了靜摩擦、庫倫摩擦和黏滯摩擦，其中靜摩擦屬於非線性函數。當CNC工具機在加工時，當其中一軸的速度正好通過零點時，因靜摩擦力帶來的暫態現象影響，使得加工精度變差。 由於目前工具機的加工曲線是在伺服馬達的位置命令下。因此，為了解決這個暫態現象問題，必需預先在軸要換象的時候補償一個值，讓這系統所產生的力矩能迅速克服靜摩擦力並到達穩態，使之加工出來的成品能維持一定的高精度。然而，由於是在位置命令下一個補償值且靜摩擦力會隨著不同的加工表面而不同，使得補償值也會隨著不同的機台而改變。要每個機台及每次加工材料的改變時都要重新人工重覆測試出適當的補償量值，是相當費時且麻煩的。本篇論文主要的目的即為讓工具機能自動找出最適當的補償量值，節省人力和成本。並經由理論推導、模擬以及實際機台驗証其可行性。 Abstract As the development in industry, time and quality are more and more important. Therefore, automatic manufacturing machines are tending to high speed and high precision. In order to eliminate tracking error, appropriate compensation value of friction is necessary for manufacture of high precision when processing a circular profile or a curved surface on the CNC machine tools. Friction includes static friction, Coulomb friction and viscous friction. Static friction is nonlinear function, and it is one of the most significant sources of nonlinear disturbance for manufacturing. The transient response that is produced by static friction will deteriorate manufacturing precision because it makes the velocity profile non-smooth at zero crossing. The nonlinear component of friction such as static friction (stiction) should be overcome so that the tracking error will be eliminated. When the X-Y tables are tracking a circular profile, quadrant glitches appear at ninety degrees intervals, i.e. the motion of one axis has a zero velocity crossing and reverses direction. The quadrant glitches due to friction brought about transient phenomenon. For solving the transient phenomenon, we will make the system to reach steady state quickly by anticipatively compensating a value when the motion of one axis has a zero velocity crossing and reverses direction. But the appropriate compensated value is difficult to get through trial and error. An algorithm is developed to find the compensated value automatically. Finally, theoretical study, simulation, and experiment are performed to verify the proposed algorithm.
