基于符号有向图的汽车涡轮增压系统故障诊断  

Diagnosis of Turbo System Fault Based on SDG

作　　者：许堃[1] 王春芳[1] 彭亚丽[1] 朱明[1] 

机构地区：[1]合肥工业大学机械与汽车工程学院,合肥230009

出　　处：《汽车科技》2011年第4期18-20,24,共4页Auto Sci-Tech

基　　金：国家科技重大专项课题(2009ZX04010-021)

摘　　要：基于符号有向图SDG(Signed Directed Graph)的传统故障诊断方法融入了定量推理的方法,提高了诊断精度,然而对不可测节点的处理大多采取等效删除以及假设状态信息扩展有效节点的方法,容易引起信息丢失和诊断效率低下的问题。结合汽车涡轮增压系统诊断实例,针对含有不可测节点的故障传播路径,提出引入模糊聚类的方法,由经验知识建立故障原因及其征兆的隶属度矩阵,结合最大隶属原则、阈值原则进行故障推理,得出有效的诊断方法。The traditional SGD-based fault diagnosis method blended the quantitative reasoning methods,which has improved the precision of the diagnostic a lot.However,during the process of dealing with the unmeasured nodes,the traditional fault diagnosis mostly takes the measure of equivalent deleting or assumption on the information extension＇s active nodes.These measures will easily cause the problems of information loss and low efficient in fault diagnosis.This paper will firstly do some research from the prospective of the unmeasured nodes＇ propagation path and propose the method of fuzzy clustering method.Then establishing the membership matrix of failure causes and symptoms from the empirical knowledge and combining the rule of maximum membership and threshold methods.Lastly,by fault reasoning,a valid fault diagnosis method will be concluded.

关 键 词：符号有向图 模糊聚类 定性定量模型 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]