基于因子隐Markov模型的旋转机械故障诊断方法的研究  

Research on Fault Diagnosis Methods for Rotating Machines Based on Factorial Hidden Markov Model

作　　者：丁启全[1] 李志农[1] 吴昭同[1] 郑时雄[2] 

机构地区：[1]浙江大学机械系,杭州310027 [2]华南理工大学机械系,广州510641

出　　处：《动力工程》2003年第4期2560-2563,共4页Power Engineering

基　　金：国家自然科学基金项目 ( 5 0 0 75 0 79)

摘　　要：针对旋转机械升降速过程非平稳、重复再现性不佳的特点 ,隐 Markov模型具有很强的针对性。因子隐 Markov模型是一种多链隐 Markov模型 ,它是隐 Markov模型的一种扩展形式。作者将因子隐 Markov模型引入到旋转机械升降速过程的故障诊断中 ,提出了基于因子隐 Markov模型的旋转机械故障诊断方法 ,并且利用它成功地对旋转机械的故障进行了分类。实验结果表明 :该方法是有效的。图 4表 2参Based on speed up and speed down(SUSD) process of rotating machinery had characteristics of big information, non stationary, worse repetition and reappearance, Hidden Markov Model adapted to deal with this process very much. FHMM had multi chains structure, and was a generalization of HMM. In this paper, FHMM was introduced to fault diagnosis in rotating machinery, fault diagnosis method based on FHMM was presented, by which, made fault classification successfully, experiment results showed that this method was very efficiency. Figs 4, tables 2 and refs 8.

关 键 词：旋转机械 因子隐Markov模型 故障诊断 模式分类 

分 类 号：TH17[机械工程—机械制造及自动化] TN911[电子电信—通信与信息系统]