本文以實驗設計法針對M1.2擠壓絲攻進行實驗，探討不同之刀具幾何參數應用於鋁合金7075的攻牙品質，並進行最佳化分析。本文實驗考慮刀具之外徑、四面寬、根徑和無效牙長為刀具幾何設計參數，利用動力計量測攻牙實驗過程之扭矩值與軸向力反應訊號，再藉由電荷放大器及擷取卡將訊號輸出到電腦，透過Labview撰寫之人機介面程式監控實驗並輸出結果。實驗後之試片用於觀察牙型輪廓和估算牙型填充率，其程序如下：首先使用金相研磨機將試片研磨至螺紋孔中心面，再以光學顯微鏡拍攝牙型後，將圖像輸出至Matlab進行二值化、檢驗牙型輪廓，並以影像處理軟體ImageJ量測牙型填充率。實驗設計採用適合二階反應曲面建構的Box-Behnken四因子三水準設計，進行28組實驗。使用統計軟體Minitab進行實驗結果之變異數分析，建立最大扭矩值與牙型填充率之回歸模型，並討論各因子對品質特性的影響，最後將此模型考量為限制最佳化問題以進行最佳化，並以最佳化結果再次進行實驗驗證，結果顯示預測模型具有足夠的準確度，期許本文能協助業界改善擠壓螺絲攻刀具之設計。;This study investigates the influence of different tool geometric features on M1.2 form tap and inspects the thread quality of form tapping on 7075 aluminum alloy. This study uses the method of design of experiment to analyze and thus optimize the torque value and the thread filling rate when form tapping. The factors of tool geometric include outer diameter, tool width, root diameter, and the chamfer (lead) length. A dynamometer is used to measure the torque and the thrust force values during tapping. The charge amplifier amplifies the voltage signal exported by dynamometer afterwards. A data acquisition card exports aforementioned analog signals to a computer. The results of the experiments are monitored and saved by the brain-machine interface code which was coded by Labview. Grinding the specimen to half section via metallography grinder, the images of the thread geometrics are taken by an optical microscope afterwards. Matlab and ImageJ are applied to draw the outline of thread geometrics and calculate the filling rate of the thread.28 sets of experiments are designed in accordance to the Box-Behnken design, which is a type of second order response surface methodology, with three levels to each of the four factors. Analysis of variance(ANOVA) based on the experiment results established the regression equation of the thread fill rate and the maximum torque by using Minitab software. According to the effects of factors on tool geometry, this study examines the optimal design for increasing thread filling rate, and reduces maximum torque in the threading process. The resulting optimization is validated by the experiments, and proves the analysis model has achieved a certain level of accuracy when used on the internal threading process. The model is expected to support an optimal tool design for the industry.
