旋转机械故障诊断中的EMD-HMM识别方法研究  

Study to EMD-HMM Recognition Approach in Malfunction Diag-nosis for Rotating Machinery

作　　者：李志农[1] 蒋静[1] 唐高松[1] 韩捷[1] 

机构地区：[1]郑州大学机械工程学院,河南郑州450001

出　　处：《矿山机械》2007年第12期117-120,共4页Mining & Processing Equipment

基　　金：国家自然科学基金(50675209);河南省杰出人才创新基金(0621000500);河南省教育厅自然科学基金(2006460005)资助项目。

摘　　要：结合经验模态分解(Empirical Mode Decomposition,EMD)和隐Markov模型(HMM)各自的优点,提出了一种基于EMD-HMM的旋转机械故障诊断新方法,该方法是先对振动信号进行EMD分解,对分解后的本征模函数进行希尔伯特变换得到其边缘谱图,并用希尔伯特边缘幅值谱作为特征向量,输入到各种故障模式的HMM中进行识别。该方法非常适合旋转机械设备非平稳运行过程中的故障诊断,实验结果表明该方法是有效的。Combining the advantage of Empirical Mode Decomposition (EMD) and Hidden Markov Model (HMM), the paper offered a new kind of malfunction diagnosis approach for rotating machinery based on EMD- HMM. In the new approach, the vibration signal was firstly conducted EMD decomposition, the decomposed Intrinsic Mode Functions (IMF) was made hilbert transformation to acquire its border spectrum, then amplitude spectrum of border spectrum served as feature vector to input into the HMM for malfunction recognition. The approach is very suitable for malfunction diagnosis during non-stationary opera- tion process of rotarying machinery. The experimental result showed that the approach was effective.

关 键 词：经验模态分解(EMD) Hilbert—Huang变换(HHT) 隐Markov模型(HMM) 故障诊断 

分 类 号：TH165.3[机械工程—机械制造及自动化]