應用於失效模式與效應分析之創新風險評估技術

Abstract

風險優先數(Risk Priority Number, RPN)為一般在失效模式與效應分析(Failure Mode and Effect Analysis, FMEA)中用於評估各失效模式重要性的準則，由各失效模式於其嚴重度(S)、發生度(O)和偵測度(D)三方面給予1-10分的評比。RPN為各失效模式所得到之評分的乘積。但在傳統的RPN方法有RPN值的重複性太高、無法比較SOD間的順序權重、且沒有考慮到各項目中的失效模式與失效原因的直接與間接的關係等問題，而無法確實排出各失效原因的相對重要性排序的問題。因此本研究提出一套方法，結合理想解接近度偏好順序排序評估法 (Technique for Order Preference by Similarity to Ideal Solution, TOPSIS) 與決策實驗室分析法 (Decision making trial and evaluation laboratory, DEMATEL)，用於FMEA中處理失效模式的排序問題。此方法先藉由TOPSIS找出SOD間的順序權重和減少RPN值重複性太高，再藉由DEMATEL來找出失效模式以及失效原因的直接與間接關係以確保能確實地排出失效模式的重要順序。本研究應用所提之方法分別於評估兩個實際案例，重新排定其失效風險的優先順序，並且與其他方法比較。研究結果發現本研究所提之方法確實可以有效解決傳統RPN方法中所發現的問題，亦得到更合理的風險評估結果提供決策者重要的資訊做決策參考。

The risk priority number (RPN) method is commonly applied in the failure mode and effect analysis (FMEA) to assess the importance of failures. RPN is the mathematical product of the three parameters of a failure mode that is rated between 1 to 10 in terms of its severity (S), occurrence (O), and detection (D), respectively. However, the RPN method has been found with the high reduplicated RPN values, unable to consider the ordered weights of SOD and failed to consider the direct and indirect relationships between the failure modes and causes of failure. This study provides a method that combines the technique for order preference by similarity to ideal solution (TOPSIS) and the method of decision making trial and evaluation laboratory (DEMATEL) to prioritize failures in FMEA. TOPSIS is used to decrease the reoccurrence of the RPN values while DEMATEL is applied to explore the direction and indirection relationship in failure modes and causes of failure and rank the most important order of failure mode indeed. In numerical verification, two FMEA cases are presented to illustrate the effectiveness of the proposed approach. Compared with the conventional RPN method and other methods, the proposed method can resolve abovementioned of RPN ranking issues and give a more appropriate risk assessment than other listed approaches to provide valuable information for the decision makers.