为了能够提取隐含在振动信号中的故障特征,利用第二代小波对称、紧支撑和冲击振荡衰减的特点,有效地提取具有冲击响应特性的故障特征。实验证明,即使采用较小支撑区间的此类小波,也可获得理想的效果。另外,为了获得与原始信号相同的时间分辨率,采用单支重构的方法分别对逼近信号和细节信号处理,得到了转子碰摩故障的时域响应特征,为故障诊断和预示提供了一种分离故障时域特征的方法。
In order to obtain fault feature that submerged in vibration signals, fault features with impact response properties are efficiently extracted by virtue of the characteristics of the second-generation wavelets following : symmetrical, compactly supported and decaying oscillation with impulsive signature. Experimental results shown that an approved effect can be achieved even if one such wavelet with less supported interval is adopted. Moreover, for the purpose of getting the same time resolution as original signal, a technique named single branch reconstruction is used to rebuild the approximation signal and the detail signal respectively. Then the time domain response features of rotor rub-impact fault are gained. It gives a new approach for separating the time domain features from vibration signal in fault diagnosis and prognosis.