摩擦振动信号的经验模式分解和多重分形研究

为提取摩擦振动的特征信号和实现摩擦振动特征信号的定量表征，在摩擦磨损试验机上进行了船用柴油机缸套-活塞环摩擦副摩擦磨损模拟试验，应用经验模式分解(Empirical mode decomposition,EMD)对非线性、非平稳的摩擦振动信号进行分解，获得若干个本征模式分量(Intrinsic mode function,IMF)。根据振幅范围和互相关系数，从EMD的所有IMF中选择反映摩擦振动特征的IMF分量重新合成摩擦振动特征信号，利用改进的多重分形算法对摩擦振动特征信号进行多重分形分析，得到摩擦振动信号的多重分形谱，进一步求取摩擦振动信号多重分形谱的宽度、极大值、维差以及非对称指数。研究结果表明，经验模式分解能够实现微弱摩擦振动特征信号的提取，多重分形谱参数可以作为摩擦振动信号的特征值。

Study on Empirical Mode Decomposition and Multifractal of Frictional Vibration Signal

For the purpose of extracting and quantitatively characterizing the characteristic signal of frictional vibration, simulation experiments of friction and wear of a piston ring against a cylinder in marine diesel engine were conducted on the testing machine. The nonlinear and nonstationary frictional vibration signals were decomposed and several intrinsic mode functions (IMFs) were acquired by using empirical mode decomposition (EMD). Based on the range of amplitudes and the crosscorrelation coefficient, the IMFs involving the characteristic of frictional vibration were extracted to resynthesize the characteristic signal of frictional vibration which was then analyzed by utilizing the improved multifractal algorithm to derive the multifractal spectrum as well as its width, peak value, dimensional difference and asymmetric index. It was shown by the results that the weak characteristic signal of frictional vibration could be extracted by EMD and the frictional vibration signal could be characterized by the parameters of the multifractal spectrum.