基于连续交叉小波相干分析和自适应CYCBD的轴承故障诊断

Bearing fault diagnosis based on continuous cross wavelet coherence analysis and adaptive CYCBD

最大二阶循环平稳指标盲解卷积(maximum second-order cyclostationarity blind deconvolution,CYCBD)能从强背景噪声信号中恢复周期脉冲,是轴承故障诊断的有效方法。故障特征频率是CYCBD的关键参数,由于滚动轴承存在制造误差、滚子滑移等现象,导致真实的故障特征频率与理论值存在偏差,降低了CYCBD的有效性。同时,故障轴承测试信号中含有大量噪声和谐波干扰,也降低了CYCBD的故障特征提取能力。对此,提出了一种基于连续交叉小波相干分析和自适应CYCBD的轴承故障诊断方法,首先,利用正常轴承、故障轴承测试信号的交叉小波相干分析获取轴承故障共振频带。其次,基于3种归一化的周期检测指标提出一种新的周期检测技术以获取真实的轴承故障特征频率。最后,基于轴承故障共振频带信号和真实轴承故障特征频率进行CYCBD滤波,并针对滤波信号进行Teager能量算子解调分析得到能量频谱,从而进行轴承故障诊断。仿真信号和高速列车牵引电机轴承试验信号的分析结果表明,该方法能够有效识别轴承故障特征,且优于传统的CYCBD方法。

Maximum second-order cyclo-stationarity blind deconvolution(CYCBD) method can recover periodic impulses from strong background noise signals,and it is an effective method for bearing fault diagnosis.Bearing fault feature frequencies are key parameters of CYCBD,due to manufacturing errors and roller slippage in rolling bearings,actual fault feature frequencies and theoretical values have deviations to reduce the effectiveness of CYCBD.Meanwhile,testing signals of faulty bearing contain a large amount of noise and harmonic interferences to also reduce fault feature extraction ability of CYCBD.Here,a bearing fault diagnosis method based on continuous cross wavelet coherence analysis and adaptive CYCBD was proposed.Firstly,the cross-wavelet coherence analysis of testing signals from normal and faulty bearings was used to obtain bearing fault resonance bands.Secondly,a new periodic detection technique based on 3 normalized periodic detection indexes was proposed to obtain actual bearing fault feature frequencies.Finally,based on the obtained bearing fault resonance band signals and actual bearing fault feature frequencies,CYCBD filtering was performed,and Teager energy operator demodulation analysis was performed for the filtered signals to obtain the energy spectrum for bearing fault diagnosis.The analysis results of simulated signals and test signals of high-speed train traction motor bearings showed that the proposed method can effectively detect bearing fault features;it is superior to the traditional CYCBD method.