基于双调Q小波变换的瞬态成分提取及轴承故障诊断应用研究

Transient feature extraction based on double-TQWT and its application in bearing fault diagnosis

因轴承的剥落、裂纹等局部故障易致运行时振动信号中出现瞬态成分,而轴承故障振动信号为非平稳信号,含高、低振荡成分,传统的线性信号处理方法及基于频率的分解方法均存在一定局限性。对此,研究基于信号振荡特征而非频率特征的双调Q小波变换,设定不同Q因子小波将轴承故障信号非线性分解成低、高振荡及噪声成分,轴承故障瞬态成分对应低振荡成分,提取低振荡成分即能实现轴承故障瞬态成分提取。通过轴承故障状态下瞬态成分检测表明,该方法能有效提取轴承故障瞬态成分。经与均值滤波、小波阈值及经验模态分解(EMD)等方法比较,验证该方法的优越性。

Local faults in rotating machinery bearings are easy to cause transient impulse response components m vibration signals. In order to realize bearing fault diagnosis under strong noise conditions, it is crucial to extract fault features from vibration signals. But a bearing fault vibration signal is a non-stationary one, it consists of high and low resonance components, traditional linear methods and signal decomposition methods based on frequency have certain limitations. To overcome these limitations, a nonlinear signal analysis method named double tunable Q-factor wavelet transformation （double-TQWT） was proposed, it was based on signal resonance characteristics rather than frequency features. By using the double-TQWT, a bearing fault vibration signal was decomposed into high and low resonance components based on different resonance characteristics. The bearing fault transient component had a low Q-factor and was decomposed into low resonance components. Extracting these low resonance components could realize extracting bearing fault transient components. The transient components for bearing fault signals under strong noise conditions were extracted and analyzed. The results showed that the new method is superior to the average filtering method, the wavelet threshold algorithm, and the EMD.