MRSVD-EMD方法在滚动轴承故障诊断中的应用

Application of MRSVD-EMD in the Fault Diagnosis of Rolling Bearings

经验模态分解(EMD)广泛应用在故障分析过程中,特征提取时从状态信息中提取与机械设备故障有关的信息[1]。针对经验模态分解受噪声影响较大的问题,提出多分辨奇异值分解的方法,可以先利用多分辨奇异值分解将信号分成具有不同分辨率的近似信号和细节信号实现信号降噪,再进行经验模态分解,并计算其Hilbert边际谱得到准确的特征频率。实验通过仿真信号和滚动轴承故障特征提取,证明了多分辨奇异值分解(MRSVD-EMD)方法在滚动轴承故障诊断中能有效去除信号中的噪声成分,提取故障特征频率。

Empirical mode decomposition（EMD）is widely employed in fault analysis and it extracts information related to mechanical failure from status information of feature extraction[1].In view of the fact that EMD is greatly affected by noise,it is proposed to apply the method of MRSVD-EMD by using Multi-resolution Singular Value Decomposition（MRSVD）to divide the signal into approximate signals and detail signals with different resolutions,so as to realize the noise reduction of the signals;accurate characteristic frequency is achieved by the Hilbert marginal spectrum calculation.It is proved that the MRSVDEMD method can effectively remove the noise component in the rolling bearing fault diagnosis and extract the fault characteristic frequency by the simulation signal and the rolling bearing fault feature.