基于MP稀疏分解与空间点群的机械故障模式表征方法  

A mechanical fault pattern characterization method based on MP sparse decomposition and spatial point group

作　　者：程卫东[1] 尹尧心 CHENG Weidong;YIN Yaoxin(School of Mechanical,Electronic and Control Engineering,Beijing Jiaotong University,Beijing 100044,China)

机构地区：[1]北京交通大学机械电子与控制工程学院,北京100044

出　　处：《北京交通大学学报》2020年第1期98-105,共8页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：国家自然科学基金(51275030);国家重点研发计划(2016YFB1200601-B24)。

摘　　要：机械系统产生的振动信号是由多激励力源产生的振动响应的混合,为了实现某一机械故障源和其他振动源的分离,需要对机械故障模式特征进行表征.针对振动信号时域表征难以区分机械故障模式特征的问题,将空间点群引入到故障诊断领域,提出一种具有明显物理特征的基于MP稀疏分解与空间点群的机械故障模式表征方法.以滚动轴承内、外圈故障实测振动信号为实验对象,首先对滚动轴承故障信号进行分段切片化处理.然后对每段切片信号进行MP稀疏分解,将之分解为少数五维原子的线性组合,通过残差信号能量确定分解原子个数.最后从定性和定量两方面对原子的两个维度进行选择,形成二维空间点群,对机械故障模式进行表征.通过与时域信号表征进行对比试验,结果表明该表征方法能够更准确地表征机械故障模式.The vibration signals generated by the mechanical system are a mixture of vibration responses that multiple excitation forces generate.In order to separate a mechanical fault source from other vibration sources,it is necessary to characterize the characteristics of the mechanical fault pattern.The time-domain representation of vibration signals is difficult to distinguish the characteristics of mechanical fault pattern.To solve this problem,the space point group is introduced into fault diagnosis,and a characterization method for mechanical fault pattern based on MP sparse decomposition and spatial point group with obvious physical characteristics is proposed.The vibration fault signal of inner and outer ring of rolling bearing is taken as the experimental object.Firstly,the fault signal of rolling bearing is sliced in sections,then each section is decomposed into a few linear combinations of five-dimensional atoms by MP sparse decomposition.The number of atoms after decomposition is determined by residual signal energy.Finally,two dimensions of atoms are selected from qualitative and quantitative perspectives,and a two-dimensional spatial point group is formed to characterize the mechanical failure pattern.According to the experimental results,compared with time-domain signal characterization,the proposed method can characterize the mechanical fault pattern more accurately.

关 键 词：故障诊断 分源监测 机械故障模式表征 稀疏分解 空间点群 

分 类 号：TH165.3[机械工程—机械制造及自动化]