匹配追踪过程复合熵矩转子故障诊断

Research on Rotor Fault Diagnosis Using Matching Pursuit Decomposition and Composite Entropy Moments

转子系统作为大型机械的核心部件工作环境十分复杂,故障种类多样且其振动信号包含大量噪声,所以特征向量难以有效提取。为此,利用匹配追踪分解对信号进行降噪,然后提取信号的奇异谱熵和功率谱熵作为故障特征,并提出复合熵矩的概念,最终利用复合熵差矩的均值和方差实现对转子故障的诊断和识别。在试验台上模拟并采集四种转子常见的有效故障信号,并以不平衡故障作为目标故障为例进行验证,根据均值和方差最小实现准确诊断,实验结果证明该方法的有效性和实用性。

As the core components of large machinery, the rotors have complex working environments and conditions. Their fault types are various and their vibration signals contain lots of noises. So, the characteristic vectors of the signals may not be extracted readily and effectively. In this paper, the method of matching pursuit decomposition was used to reduce the noise of the signals. The signal singular spectrum entropy and power spectrum entropy were extracted as the fault feature. The concept of composite entropy moment was put forward. Finally, the mean value and the square root difference of the composite entropy moments were used to realize the diagnosis and identification of the rotor’s faults. The effective fault signals of four kinds of rotors were simulated and acquired on the test rig. As an example, the imbalance fault was used as a target fault to verify the validity of this method. The results show that the method is effective and practical for fault identification.