摘要
针对滚动轴承发生点蚀故障时峭度和1倍频能量都会发生变化的情况,提出了基于局部均值对故障信号进行分解的方法。该方法取峭度值较大有效分量信号进行重构,再做切片双谱进一步降低高斯噪声对特征频率的影响,提取故障特征1倍频、2倍频和3倍频并进行归一化处理得到特征向量。利用支持向量机对提取的特征向量进行训练与测试,从而识别故障与否及发生点蚀故障的程度。通过对实测滚动轴承振动信号的分析与诊断,验证了该方法的有效性,说明其具有良好的应用前景。
Both the kurtosis and characteristic frequency energy of the vibration signal change when rolling bearing fault occurs. Rolling bearing fault diagnosis method based on Local Mean Decomposition( LMD) to decompose the fault signal is proposed. It uses the components with higher kurtosis value to reconstruct,and then makes slice bispectrum to further reduce the effect of Gaussian noise. Eigenvector can be calculated by normalization process of various frequencies of the fault feature. Support vector machine can be used to train and test that eigenvector,so that it can recognize the fault of rolling bearing and the degree of the fault. Analysis and diagnosis of the vibration signal of rolling bearing proved the validity of the method and good prospect for its application.
出处
《华北电力大学学报(自然科学版)》
CAS
北大核心
2016年第5期62-67,82,共7页
Journal of North China Electric Power University:Natural Science Edition
基金
国家自然科学基金资助项目(51276059)
中央高校基本科研业务专项资金资助项目(2015XS25)
关键词
滚动轴承
局部均值分解
切片双谱
支持向量机
故障诊断
rolling bearing
local mean decomposition(LMD)
slice bispectrum
support vector machine
fault diagnosis