本文藉由液壓成形(Hydroforming)的方式進行最佳化引伸加工(Deep-Drawing)之研究，同時探討引伸成品之長度與精度問題。本論文的研究方式係利用田口品質工程法(Taguchi Method)，並配合有限元素模擬軟體(DEFORM-3D)進行模擬分析，來探討影響引伸加工之可控制參數的相互關係。再利用類神經網路(Artificial Neural Network)將模擬完成之數據建立資料庫，結合基因演算法(Genetic Algorithm)預測最佳製程參數。 因此，本實驗方法能在預先開模前就瞭解變形的趨勢所在，以自行建構設計之靜水壓成形模具與機台，進行實驗並將所得結果加以分析。本實驗方法經田口最佳化後之製程參數組合(上模圓角半徑、下模圓角半徑、靜水壓力)透過基因演算法重新預測之值與實驗值相比對其結果吻合。 除此之外，在不同的靜水壓下，其引伸長度將隨靜水壓增高而變高，相對之下，引伸後所產生的耳側(Earing)與皺摺(Wrinkling)，也會隨靜水壓增高而降低。 Hydro-deep drawing processes were studied on the optimization considerations of Neural Genetic Hybrid algorithm. In this research, the problems of drawing length and the finished product accuracy were discussion. To make an optimal choice of drawing parameters, the finite element method with the package DEFORM-3D were used. Consequently, an optimization die obtained form the Taguchi Method were designed and manufactured. The data bases used artificial neural network and genetic algorithm to prediction the optimization experimental parameters. It was shown that the results obtained by this method were good coincidence with experiments. In addition, it was worthy to note that the height of drawing length were directly proportional to the hydraulic pressure on back of specimen and contrary to the height of earring and wrinkling.