運用最大熵順序權重幾何平均創新多準則失效模式、效應與關鍵性分析決策方法

Using Maximal Entropy Ordered Weighted Geometric Averaging Operators in an Innovative Decision Making Approach for Multicriteria Failure Mode, Effects and Criticality Analysis

廖敏利(Ming-Li Liao)

指導教授 ： 周永燦 ； 楊康宏

中原大學/電機資訊學院/工業與系統工程研究所/博士(2016年)

https://doi.org/10.6840/cycu201600155

摘要

失效模式、效應與關鍵性分析（Failure Mode, Effects and Criticality Analysis; FMECA）是一種被廣泛使用的設計技術和分析程序，可有效提高產品的品質、安全性和可靠度。現行FMECA的分析過程是依據美國國防部編定的軍用規範MIL-STD-1629A執行FMECA程序，其分兩階段執行失效模式與效應分析(Failure Mode and Effects Analysis; FMEA)及關鍵性分析(Criticality Analysis; CA)，主要目的係在設計階段優先考慮產品失效模式，並進行改正措施，再依據關鍵性分析(CA)結果，消除產品的潛在風險。然而，傳統的風險優先數(Risk Priority Number; RPN)及關鍵性分析(CA)兩種分析方法均存在著數學計算上的誤導。雖於1993年美國國防部發布 FMECA 手冊說明如何使用MIL-STD-1629A，但1998年仍通知停用MIL-STD-1629A。2006年美國陸軍發布技術手冊 TM 5-698-4取代MIL-STD-1629A，但TM 5-698-4的RPN及CA之計算程序仍存在衡量尺度及未考量權重等誤導之問題，且此數學計算方式仍廣泛被各軍用規範、技術手冊、作業規定及教科書所引用。本研究提出運用最大熵順序權重幾何平均(Maximal Entropy Ordered Weighted Geometric Averaging; ME-OWGA)多準則失效模式、效應與關鍵性分析決策方法(Multicriteria ME-OWGA FMECA)，解決RPN及CA計算上衡量尺度與權重問題。可靠度設計人員可依據分析需求及可獲得足夠數據資料之程度，運用情境參數( Sp ) OWGA所獲得的最佳權重(W)重新排序系統潛在失效的RPN與關鍵數(Criticality Number; CN)，可直接執行一階段ME-OWGA FMECA分析，或以兩階段依序執行ME-OWGA RPN後再執行ME-OWGA CA分析。另可靠度設計者設定可靠度評估值時，常遇到無法獲得足夠精確數值資訊之技術困難，且一個經過評定的可靠度評估值無法符合實際作業需求，本研究運用情境參數ME-OWGA提出模糊失效模式、效應與關鍵性分析方法(Fuzzy ME-OWGA FMECA)，提供可靠度設計者另外一項評估設計選擇。本研究分別以戰鬥機之通訊機系統、剎車系統及冷卻系統(離心泵)三個案例，進行分析與驗證，結果顯示所提出之方法能夠準確且及時有效地計算出正確的風險優先數及關鍵性分析評估值。本研究提出之Multicriteria ME-OWGA FMECA及Fuzzy ME-OWGA FMECA兩種分析決策方法，可提供軍用及商用領域使用，有效輔助管理者在FMECA設計和分析過程中，準確決定企業資源的合理分配。

關鍵字

失效模式效應與關鍵性分析、最大熵順序權重平均、最大熵順序權重幾何平均、風險優先數、決策實驗室分析法

並列摘要

Failure mode, effects and criticality analysis (FMECA) is a widely useful design technique and powerful engineering procedure for enhancing product quality, safety and reliability. It can be applied during the initial product or equipment design phase. Most of the current FMECA procedure is in accordance with MIL-STD-1629A and FMECA guide by which to conduct the failure mode and effects analysis (FMEA) and criticality analysis (CA). These are able to prioritize the failure modes and undertake limited corrective actions toward eliminating the potential risks of the product in design or manufacturing process. The Department of the United State's Army published technical manual TM 5-698-4 to create a new FMECA method, since the formal military specifications and standards MIL-STD-1629A was cancelled. However, the product design processes of the 2-phases do not take appropriate corrective measures to eliminate the product risks in the first design phase. Thus, the risk priority number (RPN) and criticality analysis (CA) methods have serious flaws from a technical perspective which are still widely used and misleading in the related FMECA handbook and instruction manual. To resolve these problems, this study proposes the multicriteria ME-OWGA FMECA, which is based on the knowledge-based to obtain the different factor criteria and uses the maximal entropy ordered weighted geometric averaging (ME-OWGA) operators to obtain the optimal weighting vector and calculates the FMECA values for a system or subsystems in 1-phase and moreover circumvents the difficulties of mathematical operators. This study provides two options to conduct the FMECA evaluation, which the 1-phase ME-OWA FMECA method can be selected, if a product has the safety or risk concerns, and the operational environment and available data are sufficient. Otherwise, the 2-phases ME-OWGA RPN and ME-OWGA CA methods can be selected. Furthermore, the crisp values of reliability analysis from a technical perspective that may be not realistic in real applications. This study proposes a fuzzy ME-OWGA FMECA method, which is evaluated in fuzzy linguistic terms. Based on the knowledge-based parameters, to obtain the weighting factors to compute the fuzzy ME-OWGA FMECA number for subsystems during product design. A fighter communication system, braking system and cooling system (centrifugal pump) were used as case studies to illustrate the proposed method, which compares the proposed two approaches ME-OWGA RPN and ME-OWGA CA with the RPN and CA methods, which finally compares the ME-OWA FMECA with FMECA. The results indicate that the proposed multicriteria ME-OWGA FMECA method can efficiently shortened the design process and more accurately calculate than the MIL-STD-1629A, FMECA guide, TM 5-698-4, FMECA handbook and other equivalent instruction manuals. The experimental results demonstrate that the proposed method provides the both accurate and discriminating analysis information which helps decision making in the product design processes.

並列關鍵字

Failure Mode Effects and Criticality Analysis (FMECA) ； Maximal Entropy Ordered Weighted Averaging (ME-OWA) ； Maximal Entropy Ordered Weighted Geometric Averaging (ME-OWGA) ； Risk Priority Number (RPN) ； Decision Making Trial and Evaluation Laboratory (DEMATEL)

參考文獻

1. Aczel, J. and Saaty, T.L. (1983) ‘Procedures for synthesizing ratio judgments’, Journal of Mathematical Psychology, Vol. 27, No. 1, pp. 93-102.

2. Adamantios, M. (2000) ‘Reliability allocation and optimization for complex systems’, Proceedings annual reliability and maintainability symposium, pp. 216-221.

3. Agarwala, A.S. (1990) ‘Shortcoming in MIL-STD-1629A guidelines for criticality analysis’, Processing Annual Reliability and Maintainability Symposium, pp. 494-496.

7. Bowles, J.B. (2003) ‘An assessment of RPN prioritization in a failure mode effects and criticality analysis’, Processing Annual Reliability and Maintainability Symposium, pp. 380-386.

10. Chang, K.H., Cheng, C.H. and Chang, Y.C. (2008) ‘Reliability assessment of an aircraft propulsion system using IFS and OWA tree’, Engineering Optimization, Vol. 40, No. 10, pp. 907-921.