一种复杂设备故障诊断的不确定性推理策略  

Reasoning strategies in uncertain situations for fault diagnosis of complex device

作　　者：赖于树[1,2] 李迅波[1] 

机构地区：[1]电子科技大学机电工程学院,成都611731 [2]重庆三峡学院物理与电子工程学院,重庆404000

出　　处：《电子测量与仪器学报》2011年第2期117-123,共7页Journal of Electronic Measurement and Instrumentation

基　　金：国防科技重点实验室基金(编号:51440020104DZ0203)资助项目;重庆市科委自然科学基金(编号:CSTC;2010BB7316)资助项目

摘　　要：一种有效的不确定性推理策略被提出用于提高复杂设备故障诊断的效率。首先,针对某些复杂设备很难有效地创建基于元器件层次分解的诊断树的特点,提出了基于故障类别信息的分层级诊断推理策略;然后,在不同诊断层级中,则采用基于模糊多属性群决策的方法来确定各规则的检索优先权,以进一步提高规则检索的效率。基于这种推理策略的故障诊断实例验证了它的有效性与高性能:可以有效提高诊断结论准确性,并减少人机交互次数,提高诊断推理的效率。A novel reasoning strategy in uncertain situations is proposed to enhance the diagnostic efficiency for complex devices. According to the characteristics of complex device faults, the hybrid inference method combined with rule-based reasoning and case-based reasoning is adopted. In the rule-based reasoning, an efficient retrieving method is crucial. The usual retrieving method is based on a diagnostic tree built according to the hierarchical decomposition of devices, but it＇s difficult to achieve such diagnostic tree for some complex devices fault diagnosis. So a multiple hierarchical diagnosis method based on fault categories is proposed. And then, the retrieving priorities of rules in a certain diagnostic level are determined based on a fuzzy multi-attribute group decision making method, which can more effectively work out the priority factors. Based on this reasoning strategy, a fault diagnosis system for intelligent instruments was developed in VS.net and SQL2005. The high performance of this diagnosis system validated the reasoning strategy.

关 键 词：不确定性推理 分层级诊断 模糊多属性群决策 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]