五軸CNC 切銷具有許多優點，然而五軸刀具路徑之產生遠較三軸CNC 刀具路徑之產生複雜許多。主要是因為五軸CNC 據有兩個旋轉軸，因此刀具與工件之間的幾何關係變得非常複雜。而且在同一個刀具接觸點(cutter contact point) 上，刀軸有無限的選擇，進而其刀具夾具也對應地有無限的位置。因此該如何選擇最好的刀具軸向與位置，便成為一必須探討的問題。此研究的目的即是在所有可能的刀具軸向與位置中，選擇一組最佳的軸向及位置。由於在兩個刀具接觸點間，如果有過大的刀具軸向變化，不僅造成機器需要較大的馬力，同時也常常在工件上產生挖槽，造成切銷誤差過大，因此需要將刀軸的變化盡可能地減少。同時在切銷過程中應該隨時避免刀具過切與干涉，也是一個必須考慮的因素。由於在每一刀具接觸點上所有的不會發生過切與干涉的刀軸，可以用一個椎體的形狀來包含。而沿著切銷刀路每一點都對應一個椎體，因此最佳化的刀路便會落在所有椎體的集合之中。我們便是打算利用動態規劃(dynamic programming)的方法，在所有可能的解中找到最好的刀路。最後我們也將利用本實驗室的五軸CNC 雕刻機，驗證此刀路，並與SURFCAM 計算出來的刀路相互比較。 Five-axis machining has become more and more popular in mold manufacturing industry mainly because it offers better surface finishing, larger removal rates and less setup time. For all types of five-axis machines, there exist infinite choices of the admissible relative tool axis and tool holder position to the part surface at each cutter contact point. The selection of tool axis and tool position will affect the cut error and material removal, and hence need careful evaluation to determine. The main purpose of this research is to find the optimal tool path for five-axis machining among all admissible tool paths. There are three major concerns in this research: (i) to define a suitable mathematic index to represent the performance of five-axis tool paths, (ii) to develop an optimization searching algorithm for tool paths and (iii) to verify the final tool path and compare the machining results with other traditional five-axis machining algorithm (such as 「constant lead」 or 「normal to part」 algorithms). In this research, we will use different weights for tool variations, distance to avoid collision surface and the removal rate to define a performance index. Then we will use the dynamic programming technique to find the optimal tool path. Finally, we will compare the efficiency and accuracy of two workpieces machined by the five-axis engraver (as shown in Figure 1) in our laboratory. The results of this research will improve the tool path generation for five-axis machining and may be published in computer-aided manufacturing journals. It can also be applied to industrial machines. 研究期間：9308 ~ 9407
