基于广义证据理论的FMEA评分冲突融合模型  

A Rating Conflict Fusion Model of FMEA Based on Generalized Evidence Theory

作　　者：周雪莲[1] 刘枫[1] 唐永川[1] 

机构地区：[1]西南大学计算机与信息科学学院,重庆400715

出　　处：《西南大学学报（自然科学版）》2016年第12期108-114,共7页Journal of Southwest University(Natural Science Edition)

基　　金：国家科技支撑计划项目(2015BAK41B00);国家"863"计划课题(2012AA041101);重庆市集成示范计划项目(CSTC2013JCSF00004)

摘　　要：针对失效模式与后果分析中不同专家之间的评分差异,在分析其差异来源后,基于广义证据理论(GET)对专家之间的评分冲突进行合理解释和量化表示;随后,对有效量化冲突后的专家评分,并利用GET中的广义组合规则(GCR)对其进行融合;最终,根据FMEA评分的融合结果,计算出FMEA的风险顺序数.利用本文方法对航空发动机涡轮转子叶片进行FMEA分析,实验数据表明本文方法比基于经典证据理论的FMEA冲突融合方法更能精确地表达专家之间的评分冲突.In order to handle the rating conflicts among each expert in failure mode and effects analysis （FMEA）, after analyzing the source of conflict, the rating conflict in FMEA is interpreted and expressed properly based on generalized evidence theory（GET）. Then, the ratings along with the quantified conflict of different experts are combined by generalized combination rule（GCR） in GET. Finally, the risk priority number（RPN） of FMEA is calculated based on combined ratings. The proposed method is used in a real case study to the rotor blades of an aircraft turbine, and compared with the method of conventional Demp- ster-Shafer evidence theory, it can express the rating conflict among experts more accurately.

关 键 词：失效模式与后果分析(FMEA) 广义证据理论 广义组合规则 风险顺序数(RPN) 冲突融合 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]