基于CS和MOMEDA的滚动轴承故障特征提取

Fault Feature Extraction of Rolling Bearing Based on CS and MOMEDA

滚动轴承发生故障时会产生周期性脉冲,在噪声干扰下微弱特征难以提取且运算效率低。应用多点优化最小熵解卷积修正(MOMEDA)方法提取故障周期,增强周期性脉冲信号,但在实际运用中该方法提取故障周期的运算效率较低。应用压缩感知(CS)方法对原始信号进行预先处理,通过稀疏表示以及正交匹配追踪算法(OMP)信号重构达到降噪目的。通过试验验证MOMEDA较之其他方法的优越性,CS方法对前者运算效率的提升具有明显效果。

Fault bearing generates periodic impulses,and it is difficult to extract weak features and the computational efficiency is low under noise interference.In this paper,it firstly extracts fault period and enhances periodic impulsive signal with MOMEDA(multi-point optimization minimum entropy deconvolution adjusted)method,but the computational efficiency of this method is low in practical application.Then,it pre-treats the original signal with CS(compressive sensing)method,and achieves noise reduction with sparse representation and OMP(orthogonal matching pursuit)signal reconstruction.The experiments show that MOMEDA is superior to other methods,and CS method has obvious effect on improving the operation efficiency of the former.