经验模态分解端点波形延拓改进方法研究

为提高经验模态分解(EMD)的分解精度,针对分解过程中存在的端点效应问题,在对已有波形匹配延拓方法改进的基础上,提出了一种改进后的基于端点波形匹配的经验模态分解方法。通过对齿轮裂纹故障模拟试验结果分析表明,当齿轮转速为600r/min时,5 120个实验数据分解时间约为1.2 s,所得第6个IMF分量的幅值谱峰值频率为10 Hz,对应着系统的回转频率。第2个IMF分量的峰值频率为550 Hz,对应着系统的啮合频率。所提出的方法能够分解出齿轮系统的特征分量,具有较高的分解精度和计算效率。

Research on the improved method of the endpoint wave extension of empirical mode decomposition

To improve the decomposition precision of the empirical mode decomposition（EMD）,for the end ef-fect produced in the decomposition process of empirical mode decomposition,a new improved method of empirical mode decomposition is founded using a new cut off condition during the process of the decomposition.Through the analysis on the signal produced by the experiment with the crack fault simulation,the results indicate that the decomposi-tion time for the 5 120 groups of experimental data is about 1.2 s.The amplitude of spectral peak frequency of the sixth IMF components is 10 Hz,that is corresponding to the rotation frequency of the system.And the peak frequency of the second IMF is 550 Hz,that is corresponding to the meshing frequency of the system.With the proposed method,the char-acteristics components of the gear system can be decomposed,and it has the high precision and computational efficiency.