應用VIKOR與因素分析於多重品質特性製程之改善

Abstract

隨著產品或製程複雜度的增加，工業界為了提升產品品質及降低製造成本，往往在設計階段即考慮利用一些線外(off-line)品管最佳化技術，如實驗設計(Design of Experiment, DOE)或田口方法(Taguchi method)，來找出製程或產品參數之最佳水準設定值；為了確保後續製程的穩定，在生產階段也會利用線上(on-line)品質監控程序進行製程管制。此外，由於產品功能愈來愈複雜，品質好壞已非單一品質特性能決定，而是由多個品質特性所影響，此現象使得線外及線上品質管理工作變得非常複雜。現有之多品質最佳化文獻中並沒有考慮到各個品質特性間之品質損失值的變異，以致於所建議之方法產生的最佳因子水準組合可能使得即使整體之品質損失值不大但各個品質特性間之品質損失值的變異過大的問題。在多品質監控技術方面，雖然Hotelling T2管制圖是業界最常使用的多變量製程監控技術，然而當利用Hotelling T2管制圖監控相關性複雜或個數過多的品質變數時，會產生檢測異常點效率變差、計算負荷過大、甚至無法求算T2指標等問題。因此，本研究之主要目的是針對多品質最佳化與製程監控程序的問題，利用多準則決策(Multicriteria decision making， MCDM)中之多準則最佳化妥協解(VlseKriterijumska Optimizacija I Kompromisno Resenje in Serbian, means Multicriteria Optimization and Compromise Solution，簡稱VIKOR)，發展出一套合理有效的多品質最佳化的分析程序，並且結合因素分析(factor analysis)與VIKOR方法，構建出一套有效的多品質製程的監控程序，可改善Hotelling T2管制圖的諸多缺點，有利業界之推廣及應用。本研究最後利用三個半導體實際案例以及一個模擬案例，說明本研究方法確實有效可行。

Multiple quality characteristics should be simultaneously considered to enhance the product quality and reduce the cost as the modern products or process designs become complex increasingly. Therefore, engineers optimize the multi-response processes using Design of Experiment (DOE) or Taguchi method in the design stage and monitor the processes using multivariate control charts to assure the multivariate processes are in control. Although several procedures for optimizing multi-response processes have been developed in recent years, the associated quality measurement indices do not consider variations in the relative quality losses of multiple responses. These procedures may therefore result in an optimization in which quality losses associated with a few responses are very small but those associated with others are very large, even if the overall average quality loss is small. As for the multivariate control technique, Hotelling T2 control chart is a popular multivariate control chart in industry. However, the calculation loading of T2 index is large when monitoring variables are larger. Besides, T2 index cannot be calculated if variables number is larger than cases number. It is also hard to detect the out of control situation of process in utilizing the Hotelling T2 control chart when the relationships among process parameters are complex. In this study, we will apply the VlseKriterijumska Optimizacija I Kompromisno Resenje in Serbian, means Multicriteria Optimization and Compromise Solution (VIKOR) method and factor analysis to solve the above problem. Finally three real cases and one simulation case study are utilized to verify the effectiveness of the proposed procedure.