基于峭度原则的EEMD-MCKD的柔性薄壁轴承故障特征提取

谐波减速器用柔性薄壁轴承运行过程中因内圈长、短轴交替产生周期性冲击成分。当柔性薄壁轴承发生故障后,这种正常的周期性冲击成分和因故障引起的冲击叠加在一起,使得其故障特征提取难度很大。针对这一特点,提出基于峭度原则的EEMD-MCKD的柔性薄壁轴承故障特征提取方法。首先使用集成经验模态分解算法(EEMD)对信号进行预处理,选用峭度原则滤除信号中的无关分量和冗余分量,重构筛选后的固有模态分量(IMF)得到EEMD重构信号;在此基础上,针对柔性薄壁轴承振动信号特点进行MCKD算法进行参数优化,利用参数优化后的MCKD对EEMD重构信号进行提取。运用此方法对实测柔性薄壁轴承外圈故障振动信号进行特征提取,结果表明,准确提取到了清晰的故障特征频率。将提取效果与单一EEMD算法和MCKD算法进行对比分析,EEMD-MCKD算法提取效果更佳。

EEMD-MCKD fault feature extraction method for flexible thin-wall bearing based on kurtosis principle

During operation of a flexible thin-walled bearing used in harmonic reducer,periodic impact components are produced due to long and short shafts of inner ring alternating.When a fault occurs in the bearing,normal periodic impact components and impact caused by fault are superimposed together to make the fault feature extraction difficult.Here,aiming at this characteristic,the EEMD-MCKD fault feature extraction method based on kurtosis principle for flexible thin-walled bearing was proposed.Firstly,the fault signal was pre-processed with the ensemble empirical mode decomposition(EEMD)algorithm.Irrelevant and redundant components in signal was filtered with the kurtosis principle,and the selected intrinsic mode function(IMFs)were used to obtain EEMD reconstructed signal.Then according to characteristics of flexible thin-wall bearing’s vibration signal,the parameter optimization was done for the maximum correlated kurtosis decomposition(MCKD).Finally,the parameter optimized MCKD was used to do fault feature extraction from the EEMD reconstructed signal.The proposed method was used to extract fault features in actually measured vibration signals of flexible thin-wall bearing’s outer ring.Results showed that clear fault feature frequency was extracted in vibration signal of outer ring of flexible thin-wall bearing with the proposed method;compared with the single EEMD and MCKD algorithms,the EEMD-MCKD algorithm has a better fault feature extraction effect.