田口方法是一種工程的程序，藉由降低各種可能的變異對一產品或製程的影響，進而提昇品質。然而，田口方法在某些情況是不太有效的，例如，其僅處理間斷型（discrete ）的控制因子，導致所求結果不是最佳解。 本文提出另一有效的方式，以類神經網路求解田口動態特性問題。本文發展兩個程序：程序1可求解間斷型控制因子之最適條件，程序2可決定連續型控制因子之最佳值。利用兩個數量例子，所發展之程序與田口方法進行比較，結果顯示本文所發展之程序是有效的。

　　　　　Taguchi method attempts to enhance the quality of products and processes by minimizing the impact of the causes of variation. However, Taguchi's approach is inefficient in some cases. For instance, it only addresses the discrete control factor, leading to a departure from the real optimal solution. This study presents an effective means of applying neural networks to resolve Taguchi's dynamic problems. Two procedures are developed. Procedure 1 can determine the optimal condition when the control is qualitative. Procedure 2 can determine the optimal value of the continuous control factor. In addition, the proposed procedures are compared with Taguchi's approach through two numerical examples. The above comparisons demonstrate that proposed procedures are effective and feasible.