本文利用動顯函有限元素法，結合Hill的異相性降伏準則，模擬金屬板材圓杯深引伸成形，並透過適應性網路模糊推論系統(Adaptive Network Fuzzy Inference System, ANFIS)，預測達成形極限目標杯高之初始料片最佳化輪廓外形。 本文首先進行圓形料片之圓杯深引伸成形極限分析，並結合ANFIS進行圓杯深引伸達成形極限目標杯高之最佳化料片輪廓外形預測。圓形料片達成形極限之最低杯高為22.48mm，經最佳化預測之料片達成形極限之目標杯高為23.53mm，其杯高約增加4.67％；而圓形料片與最佳化料片之LDR分別為1.8670與1.9053，故料片之LDR可增加約2.05％。經數值模擬與實驗結果比較得知，本文之動顯函有限元素程式與適應性網路模糊推論系統可有效的預測圓杯深引伸達成形極限之最佳化初始料片之輪廓外形。 The topic of this thesis is to combine the dynamic-explicit FEM and Hill’s anisotropic yield criterion to simulate the deep drawing process of cylindrical cups. Besides, the adaptive network fuzzy inference system (ANFIS) is applied to predict the optimum profile of blank with the target height of the cup in the forming limit. To begin with, the forming limit of the circular blank in deep drawing of cylindrical cup was conducted. With the employ of ANFIS, the optimum profile of blank with the modified target height of cup in the forming limit then would be predicted. The lowest height of the cup of circular blank in the forming limit was 22.48mm, and the objective height of the cup of optimum blank in the forming limit was 23.53mm, so the height of the cup increased by 4.67%. The LDR of the circular blank and the optimum blank were 1.8670 and 1.9053 respectively, so the LDR of blank could increase by 2.05%. The comparing results of experiment and of simulation showed that the dynamic-explicit FEM and ANFIS predict the optimum blank of the forming limit in the deep drawing process of cylindrical cup effectively.
