基于循环双谱二次切片分析的滚动轴承故障诊断研究

Fault diagnosis of a rolling element bearing based on twice-slice of a cyclic bispectrum

发生早期微弱故障的滚动轴承,由于其故障特征频率处能量比较小,易于被噪声淹没,仅仅利用二阶循环平稳已经不能有效提取故障特征。而高阶循环统计量对噪声具有较好的免疫力。但是高阶循环统计量谱为多维谱图,图形化表示比较困难。本文对三阶循环累积量谱,即循环双谱,进行了简化,提出了循环双谱二次切片———循环双谱中心频率处切片的水平切片谱。并用此简化的循环双谱对滚动轴承早期故障做特征提取。仿真和实验结果证明了方法的有效性。

Bearing fault characteristic frequencies contain very little energy when the early fault occurs, and they are usually overwhelmed in noise. It is difficult to find them in their frequency spectra with the second order cyclic statistics analysis. To overcome this difficulty, the cyclic bispectrum （CBS） based on the third order cyclic cumulants was presented for fault diagnosis. However, the cyclic spectrum was forth-dimensional. The displaying of the cyclic spectrum was difficult. And the calculating amount was big. So, twice-slice of a CBS, a simplified approach for detecting localized faults in outer races of a rolling element bearing, was proposed here. The results of simulations and tests indicated that the twice-slice analysis for rolling element bearing early fault diagnosis is feasible and effective.