二维隐Markov模型的故障诊断方法  

Fault Diagnosis Method Based on 2-Demensional Hidden Markov Model

作　　者：叶大鹏[1] 丁启全[2] 吴昭同[2] 

机构地区：[1]福建农林大学机电工程学院,福建福州350002 [2]浙江大学机能学院,浙江杭州310027

出　　处：《控制工程》2005年第S2期191-193,共3页Control Engineering of China

基　　金：国家自然科学基金资助项目(50075079)福建省教育厅资助项目(JA04182)

摘　　要：二维隐Markov模型(2D-HMM)具有强大的时序模式分类能力,特别适合于分析非平稳、重复再现性不佳的信号。鉴于旋转机械工作过程中存在大量的非平稳信号,将2D-HMM引入到旋转机械故障诊断中,提出了一种基于2D-HMM的故障诊断方法,并通过Bently-Nevada转子试验系统验证该方法的有效性。Two dimensional hidden Markov model has powerful capability to classify time series model. It is especially good at analyzing nonstationary or worse reappearance signal. So this model is introduced for fault diagnosis in the rotating machinery field with large amounts of non-stationary signal in these machines operation. Finally, the new method was tested by experimental data collected from Bendy-Nevada rotor experimental system (Model 24755) and the result demonstrates that it is very effective to diagnose the rotor faults.

关 键 词：二维隐马尔夫模型 旋转机械 故障诊断 非平稳信号 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]