基于典型谱相关峭度图的滚动轴承故障诊断方法

摘要

In practical applications,the types of occurring faults on rolling bearing are typical and fault impact signal is cyclical.In view of the characteristics,a typical correlated kurtogram algorithm was proposed in this paper.Based on the ideas of typical kurtogram of hierarchy and selecting central frequency,the kurtosis was replaced by correlated kurtosis,not only avoiding the noise interference introduced by the wide-band demodulation,but also making full use of the periodic information of typical fault impact,and then the method located typical fault impact signal interested in the frequency range rapidly,relying on the optimal spectral correlated kurtosis.And the algorithm was applied to determing the optimal demodulation frequency band.Through the study on simulation and wheelset bearing experiment signals,the results show that the typical correlated kurtogram algorithm can locate resonant frequency band effectively and adaptively and have a great superiority in both accuracy and stability.%针对工程实际中滚动轴承发生故障的类型具有典型性和故障冲击信号具有周期性的特点,提出了一种典型谱相关峭度图算法.该算法在借鉴典型谱峭度图算法区间划分的思想基础上,将相关峭度指标代替峭度指标,不但避免了宽频带解调引入的噪声干扰,而且充分利用了典型故障冲击的周期性信息,并通过优化谱相关峭度值,快速定位典型故障冲击信号所在的频率区间,并将该算法应用于最优解调频带的确定.通过对仿真信号和轮对轴承实验信号的分析表明,该算法无论在准确性还是在稳定性方面均表现出了极大的优越性,能够有效的自适应定位共振频带.