信息缺失时基于HMM的故障诊断方法  

HMM-based fault diagnosis method to emergencies resulting from information omission

作　　者：张春良[1] 岳夏[1] 朱厚耀[1] 周超[1] 欧彪[1] 

机构地区：[1]广州大学机械与电气工程学院,广东广州510006

出　　处：《广州大学学报（自然科学版）》2013年第4期64-69,共6页Journal of Guangzhou University:Natural Science Edition

基　　金：广东省自然科学基金资助项目(S2012010009505)

摘　　要：在突发事件导致的事故中,状态监测与故障诊断系统的研究对象已经发生很大的变化,常规设计的状态监测与故障诊断系统往往很难满足要求.文章针对信息缺失提出了一种新的状态监测与故障诊断方法.该方法基于HMM模型的似然率计算过程,提出在发生信息缺失时,对缺失信号依据各模型采用最大似然率填充的方式进行故障诊断,并通过实际数据的测试实验对该方法进行了验证.验证结果表明,信号缺失时依据各模型采用最大似然率填充的方式进行故障诊断的方法是有效的.In emergencies caused by accident, the diagnostic object of fault diagnosis system has undergone great changes. The conventional designed fault diagnosis systems are difficult to accurately diagnose. Therefore, a novel fault diagnosis method to emergencies resulting from information omission was presented in this paper. This method is based on HMM likelihood calculation process, in which the missing information is complemented with maximum likelihood padding, and was validated by experimental testing of actual data. Verification results indicate that this method can effectively solve the fault diagnosis problems when information is missing.

关 键 词：故障诊断 HMM模型 信息缺失 最大似然率 

分 类 号：TH172[机械工程—机械制造及自动化]