應用灰關聯分析與雙反應曲面法最佳化多品質問題

Abstract

實驗設計(Design of Experiments,DOE)與田口方法(Taguchi Methods)是工業界最常使用的兩大線外品質管制方法，以找出最佳因子水準組合。然而，此兩方法均只能用於最佳化單一品質特性。最近幾年，消費者對於產品品質的要求複雜。產品創新、設計、品質以及製程技術等都是衡量產品價值的重要指標。雖然我們希望所有品質特性都能是最好的，但由於資源的限制，很難達到這樣的目標。因此，需要發展一套多品質同時最佳化之演算法。本研究應用灰關聯分析法(Grey Relational Analysis Method)與理想解類似度順序偏好法(Technique for Order Performance by Similarity to Ideal Solution, TOPSIS)建立各配方之理想解灰關聯度與負理想解灰關聯度之雙反應曲面，以求解最佳配方。本研究最後應用一個新竹科學園區某半導體公司蝕刻製程之實際案例，驗證了本論文所發展之演算法確實有效。

Design of Experiment and Taguchi Method are two major off-line quality control methods used by industry to find the optimal factor-level combination. However, these two methods can only optimize single response. In recent years, the required product quality becomes complex. The innovation, design, quality of a product and manufacturing process techniques are important criteria to measure the product value. Even though we want all quality characteristics to be perfect, it is hart to reach the goal owing to the limitation of resources. Thus, we need to develop a methodology to optimize multiple responses simultaneously. This study utilizes Grey Relational Analysis and Technique for Order Performance by Similarity to Ideal Solution(TOPSIS) to construct dual response surfaces for grey relational grade of ideal solution and negative ideal solution ,respectively, for every treatment of the designed experiments to find the optimal factor-level combination. A real case from a wafer manufacturing company in Science Park at Hsinchu, Taiwan is used to demonstrate the effectiveness of the proposed method.