基于多阶FRFT滤波与4阶累积量切片谱的滚动轴承早期故障特征提取

Fault Feature Extraction of Rolling Bearing Based on Multilevel FRFT Filtering and Fourth-order Cumulant Slice Spectrum

针对滚动轴承早期故障难以提取的问题,提出了一种多阶FRFT滤波与4阶累积量切片谱相结合的方法。根据轴承特征频率曲线特点,将振动信号分成若干频率近似线性变化的信号段,分别计算每段信号分数阶傅里叶变换的最佳阶次,再逐段进行滤波,从而分离出包含故障特征信息的目标分量,最后采用4阶累积量切片谱分析目标分量,提取故障特征。结果表明:多阶FRFT滤波与4阶累积量对角切片谱相结合,既能分析非稳态信号,又能有效抑制噪声干扰;多阶FRFT滤波能准确分离出目标分量,并计算滤波后目标分量的4阶累积量切片谱的能量和峰值,提取出轴承早期故障特征。

To solve the problem of extracting early fault features of rolling bearing, the paper proposes a method which com- bined multilevel Fractional Fourier Transformation （FRFT） filtering with fourth-order cumulant slice spectrum. In this method, the vibration signal was separated into several signals changing in a linear fashion according to the curve feature of rolling bearing＇ s characteristic frequency, and the best FRFT order of each segment was calculated and each segment was filtered, and then the target component with fault features information was separated. Finally, the target component was an- alyzed with fourth-order cumulant slice spectrum and the fault feature was extracted. The result shows that： combining mul- tilevel FRFT filtering with fourth-order cumulant slice spectrum can analyze non-stationary signal and restrain noise interfer- ence; multilevel FRFT filtering can exactly separate target component, calculate the energy and peak value of fourth-order cumulant slice spectrum, and extract the earlier fault feature of rolling bearing.