改进局部均值分解方法及其在往复压缩机轴承故障诊断中的应用

摘要

Reciprocating compressors are widely used in petroleum and chemical industries, but the overlarge clear-ance faults in the bearings of the compressors occur very often due to wear and tear. In this paper, considering the non-sta-tionary and nonlinearity of the vibration signals of the reciprocating compressors, an improved Local mean decomposition (LMD) method is presented for the fault feature extraction of the reciprocating compressor bearings due to the large bearing clearance. To improve the accuracy of local mean and envelope estimation, a novel envelope-curve construction method is proposed by inserting an extremum symmetrical point between two extremum points and using the Monotonic Piecewise Cu-bic Hermite Interpolation (MPCHI) instead of Cubic Spline Interpolation (CSI). Algorithm and process of the improved LMD method are given. By comparing the decomposition results of the reciprocating compressor vibration signals with dif-ferent LMD methods, the superiority of the improved LMD method is proved. Then, the overlarge bearing clearance fault is diagnosed accurately by the frequency spectrum envelope of the first PF component amplitude.%往复压缩机广泛应用于石油、化工生产行业，其滑动轴承常因磨损而出现间隙过大故障。针对往复压缩机振动信号的强非平稳和非线性特性，提出一种改进局部均值分解(LMD)方法，并将其应用于往复压缩机轴承间隙故障诊断。通过在极值点间加入极值对称点，使用单调三次Hermite插值(MPCHI)替代三次样条插值(CSI)构造包络线，提高局部均值与包络估计函数的准确性，以此提出了改进LMD方法的算法与流程。以往复压缩机轴承故障振动信号为对象，通过与不同LMD方法比较，验证改进的LMD方法的优越性，并以PF分量幅值包络频谱实现轴承间隙大故障的准确诊断。