基于多元经验模态分解的旋转机械早期故障诊断方法*

针对旋转机械早期微弱故障诊断问题,提出了基于多元经验模态分解的旋转机械早期故障诊断新方法。首先将多个加速度传感器合理布置在轴承座的关键位置,同步采集多通道振动信息;再利用多元经验模态分解同时对多通道振动信号进行自适应分解,得到一系列多元IMF分量;最后,依据峭度准则和相关系数从中选取包含故障主要信息的IMF分量进行信号重构,提取故障特征。多元经验模态分解方法克服了EMD等方法在进行多通道数据融合时缺乏理论依据的局限性。仿真信号和旋转机械故障信号的实验结果表明,该方法明显优于EEMD方法,对齿轮和滚动轴承故障的检测精度更高,可以在强背景噪声情况下更好地提取出故障冲击特征。

Rotating machinery early fault diagnosis method based on multivariate empirical mode decomposition

Aiming at the problem in early weak fault diagnosis of rotating machinery using single sensor, a new early fault diagnosis method for rotating machinery is proposed based on multivariate empirical mode decomposition. Firstly, several accelerator sensors are reasonably arranged on the key positions of the bearing housing, which synchronously acquire the multi-channel vibration information. Then, the multivariate EMD is used to adaptively decompose the multi-channel vibration signals to get a series of multivariate IMF components, from which, finally, the IMF component containing main fault information is selected according to kurtosis criterion and correlation coefficients to conduct signal reconstruction, and extract the fault feature. The multivariate empirical mode decomposition method overcomes the limitation that traditional EMD and etc. fault diagnosis methods lack theoretical basis in conducting data fusion. The experiments on simulation signal and real fault signal of rotating machinery were carried out, and the proposed method was compared with the EMD and EEMD methods. The results show that the proposed method is obviously superior to the other two methods, the detection accuracies for the rolling bearing inner ring fault and gear tooth broken fault are higher than those for the other two methods. The method can be used to extract various impact fault characteristics of rotating machinery clearly under strong background noise condition.