基于EMD-AR谱分析的数控机床主轴故障诊断方法研究∗

摘要

Aimed at the CNC machine tool spindle system noise problem, a diagnosis method was proposed based on the empirical mode decomposition(EMD)and autoregressive(AR)power spectrum analysis meth-od. The data collection and process software of CNC machine tools spindle is ECON;the data was decom-posed into a series of intrinsic mode functions ( IMF) by using EMD, the maximum amplitude of the order of IMF was extracted and averaged, the IMF which the average amplitude in large proportion of the total amplitude was selected and analyzed by the AR spectrum. Vibration frequency was found out and contrasted with Hilbert marginal spectrum and FFT. Test results show that test machine frequency was caused by motor shaft frequency as the assemble error; Besides, the comparison and analysis show that the EMD-AR spec-trum was more effective to extract fault feature.%针对数控机床主轴系统经常出现的声音异响等故障,提出了一种基于经验模态分解( EMD)和AR( auto regressive)谱分析相结合的诊断方法。对主轴箱部位使用亿恒数据采集仪进行数据采集,将测量数据进行经验模态分解,提取几个各阶本征模态函数分量( IMF )最大幅值并求平均,选取平均后的幅值占总幅值中比例较大的几阶IMF,并进行自回归谱分析,通过试验测得测试机床主轴箱的主要频率值,并与希尔伯特边际谱及FFT进行对比。结果表明测试机床故障频率与电机轴频率吻合,是由于电机轴装配偏心造成；同时通过对比分析表明EMD-AR谱估计更能有效地提取故障频率。