摘要
针对信号采样频率过低对经验模态分解造成的虚假模态等问题,提出了一种改进的算法,即在进行分解前,对原始信号进行重构,其实质是通过内插的方式来增加采样点数,从而达到增加采样频率的目的.对模拟信号的处理结果表明,该算法消除了分解过程中包络曲线的异常波动,从而抑制了分解结果中多余模态的出现,使得对模态的物理解释更加清晰.在机械信号处理中,应用该算法成功地提取出机械信号中具有明确物理意义的故障模态,从而增加了机械故障诊断的能力.
An improved method for empirical mode decomposition was presented to solve the difficulties of the false modes due to the lower sampling frequency for signals, where the signals were reconstructed before the empirical mode decomposition and the sampling points were added by the interpolation manner to increase the sampling frequency. The improved method was verified in the simulation signals processing, and the abnormal fluctuating in the envelope curves was eliminated successfully, thus, the false modes were restrained. With the aid of the proposed method, the fault modes with clearer physical senses were extracted completely from the original mechanical signals in a practical case.
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2004年第11期1199-1202,共4页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金重点资助项目(50 3 3 50 3 0 )
关键词
经验模态分解
改进算法
采样频率
故障诊断
Failure analysis
Feature extraction
Sampling
Time domain analysis
