For the modulation of vibration signal in the piston pump, a method was presented which can get fault feature based on EEMD and the smoothed energy operation separation, and the wavelet packet was used to decompose energy of frequency area in order to extract fault feature vectors. Firstly, the EEMD method was adopted to decompose some IMF functions of vibration signal for piston pump. Secondly, some key IMF functions involving main fault information were selected to achieve demodulation information by means of smoothed energy operation separation. Then the fault information in the high-frequency area was extracted. Finally, the wavelet packet was used to decompose the energy of the frequency area in order to obtain fault featyre vectors. The result shows that the method can avoid the mode mixing efficiently, extract fault information in the high-frequency area and get the frequency of fault feature.%针对液压泵振动信号出现的调制现象,提出基于集总平均经验模态分解(ensemble empirical mode decomposition,EEMD)和平滑能量算子解调相结合的方法进行解调,并运用小波包分解频带能量的方法提取了轴向柱塞泵的特征向量.首先,利用EEMD将采集到的柱塞泵振动加速度信号分解成若干个平稳的本征模函数(IMF);然后,选取包含主要故障信息的本征模函数通过能量算子解调的方法进行包络解调,从而提取振动信号在高频谐振带的包络成分;最后,运用小波包理论提取各频带的能量作为特征向量.结果表明:基于EEMD和平滑能量算子解调的方法能有效地避免模态混叠现象,提取振动信号的包络成分,成功获得各种状态下的特征向量.
展开▼