Abstract:

In order to extract the fault feature informations of mechanical equipment,a blind information extraction method  was proposed  based  on  singular value fusion.Firstly,by independent component analysis,the source signals were separated from mechanical vibration observation signals.Then the source signals were demodulated to get envelope signals and form envelope matrixes.The envelope matrixes were processed by singular value decomposition and then the singular value vectors (including singular value averages and singular value entropy) were extracted as feature information of the detected mechanical equipment. Secondly, the separating matrix was directly processed by singular value decomposition and its singular values could also be extracted as useful feature information. Lastly, the singular value vector was obtained by information fusion as final optimum feature information of the mechanical equipment. The experimental results of hydraulic gear pump indicate that this method can be applied to feature extraction of mechanical equipment effectively.

Key words: blind , source separation;independent , component , analysis;envelope analysis;singular value decomposition;feature extraction

摘要：

针对机械设备的故障特征信息提取问题，提出了基于奇异值融合的机械盲信息提取方法。首先,由机械振动测量信号分离振动源信号，并进行包络解调组成包络信号矩阵，进而进行奇异值分解,提取矩阵的奇异值均值和奇异值熵作为故障特征信息;然后,针对分离矩阵直接进行奇异值分解，提取奇异值作为故障特征信息；最后，将包络信号矩阵奇异值均值、奇异值熵和分离矩阵奇异值进行特征层信息融合作为机械设备的故障特征信息。将该方法应用于液压齿轮泵可以有效地提取机械设备盲特征信息。

关键词: 盲源分离, 独立成分分析, 包络分析, 奇异值分解, 特征提取