基于LMD近似熵与HMM的转子故障诊断方法

本资料包含pdf文件1个，下载需要1积分

A new fault diagnosis approach for rotor system was proposed based on local mean decomposi- tion（LMD） approximate entropy and hidden Markov models （HMM）. The fine localization feature of LMD and approximate entropy combined with HMM were used to identify quantify the fault type. By using LMD method, the vibration signal of the rotor systems was made as a sum of several components of a product function （PF）, in which the instantaneous frequencies should have physical meaning. The ap- proximate entropies of the first three PF components were taken as the eigenvectors of the signal and the eigenvectors were input into HMM classifier to recognize the fault type. Simulation result showed that this method could be effectively used to extract the fault characteristics, and, combined with the dynamic sta- tistical characteristics of HMM, the rotor fault type could be identified intelligently.提出一种基于局部均值模式分解（10calmeandecomposition，简称LMD）的近似熵和隐MarkOV模型（hiddenMarkovmodel，简称HMM）的转子系统故障识别新方法．利用LMD良好的局域化特性和近似熵来量化故障特征，再与HMM结合进行故障类型识别．用LMD方法将转子信号分解成若干个瞬时频率具有物理意义的乘积函数（productfunction）PF分量之和，选取转子信号的前3个PF分量的近似熵值作为信号的特征向量，将构造出的特征向量输入到HMM分类器中进行故障类型识别．仿真表明，该方法能有效地提取故障特征，结合HMM的动态统计特性可智能识别转子故障类型．