In this paper,the research considers the inner regularity of the signal,which improves the method of boundary extending based on the method of the nearest similarity distance continuation and parallel extending method.The regularity strength of signal is determined by calculatins the similarity distance between the boundary wave and any internal signal wave.When the inherent regularity is strong,the nearest similarity distance continuation method is conducted to maintain the inherent tendency to the greatest extent;while the signal is week regularity and the signal edge changes abnormally,the local information at the edges is considered only to conduct the parallel extending method.In order to prove the improved algorithm,this paper calculates the matching distance,waveform similarity coefficient and error in marginal spectrum of frequency through the simulation analysis and tests of fault diagnosis of rolling bearings of IMFs and sirulation signal component.The results demonstrate that matching distances are 33787,7.2404,7.390 7,waveform similarity coefficients are 0999 9,0.9977,0903 4 anderror in marginal spectrum in 5 Hz is 62％ by the improved algorithm.Containers among directly continuation,parallel extending method,this method can suppress end effect of EMD.This method can improve the decomposition accuracy and achieve availably fault feature extraction,which can provide the reference for fault diagnosis of machine.%在经验模态分解(Empirical mode decomposition,EMD)中,综合考虑信号内部的规律性,在深入研究最近相似距离延拓法和平行延拓法的基础上,对EMD端点延拓方法进行了改进,通过计算信号内部波段与端点波段的相似距离,确定信号内部规律性的强弱.在信号内部规律性较强的情况下,采用最近相似距离延拓法,使信号两端最大程度地反应信号内在信息;在信号内部规律较弱的情况下,考虑到端点信号发生异常的情况,采用平行延拓法在信号端点进行预测延拓.为验证该方法,本研究通过仿真分析和滚动轴承故障进行研究,研究结果表明,改进后的EMD分解得到的IMF分量与原模拟信号分量的匹配距离分别为3.378 7、7.240 4、7.390 7,波形相似系数分别为0.999 9、0.997 7、0.903 4,边际谱频率在f=5 Hz的误差为6.2％,与直接延拓法、平行延拓法相比更准确,能有效地抑制EMD端点效应.该方法能提高信号分解精度,实现对故障特征的有效提取,为机械故障诊断提供参考依据.
展开▼
