基于MF-SVD的滚动轴承振动信号故障特征提取方法研究

摘要

In terms of calculation precision,restraining boundary effect and computation time,extremum field mean mode decomposition (EMMD) has obvious advantage comparing to empirical mode decomposition (EMD) and adaptive time varying filter decomposition (ATVFD).It can effectively extract machinery vibration signal fault characteristics.As the measured field signals are often mixed lots of noise,which influences EMMD decomposition quality seriously.Aiming at the problem,a new extracting fault feature method which is based on morphological filters-singular value decomposition (MF-SVD) denoising method,combined with EMMD,is put forward.The experimental results show fault features of rolling bearing inner ring damage can be effectively and accurately extracted.It has a good calculation speed and accuracy.%极值域均值模式分解(EMMD)在抑制端点效应、算法精度、计算耗时等方面均比经验模式分解(EMD)算法和自适应时变滤波分解(ATVFD)有着明显的优势,能够有效地对旋转机械振动信号进行故障特征提取,由于现场信号通常掺杂大量噪声,严重影响了EMMD的分解精度.针对该问题,提出了基于形态滤波-奇异值(MF-SVD)的去噪方法,并将其与EMMD相结合,建立了一种新的故障特征提取方法.实验结果表明:该方法能够有效、准确地提取旋转机械滚动轴承内圈损伤的故障特征,有着良好运算速度和精确度.