无鉴相角度域滚动轴承故障诊断方法研究

Defect Diagnostics of Roller Bearing by Angle Domain without Key Phase

利用EMD将轴承的振动信号分解成单分量信号并求其对应的瞬时频率,根据频率与转速的关系估计出单分量信号的转速,结合各自的转速对单分量信号进行角度域重采样,将时域内非平稳信号转化为角度域内平稳信号,再将所有角度域信号进行重构,并对重构后的信号进行阶比分析,得到阶比谱,从该谱中可以观测信号的特征阶次进而诊断出轴承故障.利用EMD对振动信号进行分解的过程中,可用高斯拟合的方法分析出原数据的走向,将端点进行向外延拓,以此抑制EMD的端点效应.最后将该方法与传统的计算阶比跟踪算法进行比较,表明该方法不仅无需鉴相器节省硬件,且对转速没有限制,能够进行在线分析.且在滚动轴承故障诊断实例中得到的阶比谱有明显的波峰,根据波峰的位置与理论计算的阶次可判断滚动轴承的故障类型.

The EMD is used to decompose the bearing vibration signal into single-component signals and obtain the corresponding instantaneous frequency.The speed of the single-component signals is estimated according to the relationship of the frequency and the rotational speed.By angle domain resampling of the signals,the non-stationary signal in the domain is converted into a stationary signal in the angular domain,and the signals in all angle domains are reconstructed.The order ratio of the reconstructed signal is analyzed to obtain the order spectrum.From this spectrum,the order of the signals can be observed and used to diagnose bearing failure.In the process of using EMD to decompose the vibration signals,Gaussian fitting can be used to analyze the trend of the original data and extend the endpoint outward to suppress the end effect of EMD.Finally,this method is compared with the traditional calculation order tracking algorithm,which shows that the method saves not only the hardware without the phase detector,but also has no limitation to the rotation speed and can be used for online analysis.In the case of rolling bearing fault diagnosis,the order spectrum has obvious peaks.According to the position of the peak and the order of theory calculation,the fault type of the rolling bearing can be judged.