极值域均值模式分解在轴承故障诊断中的应用  

Application of Extremum Field Mean Mode Decomposition In Fault Diagnosis of Rolling Bearing

作　　者：张云鹏[1] 盖强[2] 周洋[3] 

机构地区：[1]海军大连舰艇学院研究生管理大队,辽宁大连116018 [2]海军大连舰艇学院装备自动化系,辽宁大连116018 [3]上海江南造船厂军代表室,上海201913

出　　处：《测控技术》2011年第12期119-122,共4页Measurement & Control Technology

摘　　要：为了研究滚动轴承信号的非平稳特征,提出了将局域波方法和Parzen窗概率神经网络相结合的故障诊断方法。分析了局域波时频分析中极值域均值模式分解方法的改进方法,并提出了一种筛选停止准则。对分解所得分量,提取平均瞬时频率和能量比作为故障特征向量构造神经网络,进行状态判断。通过对现场采集的滚动轴承信号进行分析,说明了该方法的正确性与实用性。研究结果验证了该方法可以用于滚动轴承的故障诊断与故障识别,为机械故障诊断提供了一种新思路。To study the non-stationary characteristic of the rolling bearing signal, local wave method and Parzen window probabilistic neural networks are proposed. A new method to improve the extremum field mean mode decomposition of the local wave method is analyzed, and a new criterion is given for the sifting process to stop. The mean of instantaneous frequency and the energy ratio is extracted from the decomposed part, which is used to make up a neural network for fault diagnosis. By using this method to analyze the rolling bearing signal, the validity and feasibility is proved, and a novel method for the fault diagnosis of roiling bearing.

关 键 词：局域波法 神经网络 故障诊断 滚动轴承 

分 类 号：TH133.33[机械工程—机械制造及自动化] TP206.3[自动化与计算机技术—检测技术与自动化装置]