基于VMD的自适应形态学在轴承故障诊断中的应用

为有效提取滚动轴承信号的特征频率,提出了基于变分模态分解(VMD)的自适应形态学的特征提取方法。首先利用VMD将目标信号分解为有限个模态信号,依据互信息法提取与原始信号相关的模态信号,将其进行求和重构;然后利用形态学对重构信号进行降噪处理,提取出滚动轴承的特征频率。针对形态学固有统计偏移和结构元素的选择问题,利用粒子群算法来优化改进的广义形态学滤波器,实现自适应滤波。通过数字仿真实验与滚动轴承故障试验分析,将其与基于经验模式分解(EMD)的自适应形态学、包络解调方法进行比较,结果表明该方法可以有效提取故障信号的特征频率。

Application of adaptivemorphology in bearing fault diagnosis based on VMD

To effectively extract characteristic frequencies of rolling bearing vibration signals, the adaptive morphology was proposed based on variational mode decomposition (VMD). VMD was used to decompose a target signal into finite modal signals firstly. Then the modal signals related to the original signal were extracted based on the mutual information method, and they were summed to reconstruct a signal. The morphologic filter was used to reduce noise from the reconstructed signal, and the rolling bearing fault feature frequencies were extracted. Aiming at problems of morphologic structural element selection and the inherent statistical deviation, the particle swarm optimization was used to adoptively optimize the improved generalized morphological filter to realize adaptive filtering. Through digital simulation test and rolling bearing fault tests, the method was compared with the adaptive morphology based on EMD and the envelope demodulation method. The results showed that this method can extract fault characteristic frequencies of rolling bearings effectively.