【摘要】 为了抑制经验模态分解(empirical mode decomposition,简称EMD)中出现的端点效应和模态混叠现象,在信号组综合经验模态分解(ensemble empirical mode decomposition,简称EEMD)的基础上,从抑制信号干扰和噪声污染影响以及三次样条函数插值拟合误差逐级传播方面,提出利用信号支持向量机(support vector machines,简称SVM)延拓改进EEMD。通过对仿真和实测信号研究,比较了EMD和EEMD的分解,提出改进的EEMD方法不仅减少了虚假模态分量、避免了模态混叠,而且有效抑制了端点效应。与基于镜像延拓改进的EEMD方法比较表明,本研究方法的时频谱更加清晰,虚假模态分量更少,有效解决了端点效应引起的分解失真问题。更多还原

【Abstract】 It has a serious effect to mechanical fault diagnosis in accuracy that the endpoint effect and modal aliasing phenomenon appear in EMD(empirical mode decomposition).In order to restrain endpoint effect and modal aliasing phenomenon,improved EEMD(ensemble empirical mode decomposition) is put forward based on EEMD by using SVM(support vector machines) signal data extending.Compared with EMD and EEMD,the improved EEMD by using SVM signal data extending in simulation and measurement signals research results shows that not only the false modal component is reduced,and the endpoint effect is effectively restrained.And compared with the improved EEMD based on the mirror image signal extending,the improved EEMD by using SVM signal extending has more clearly time-frequency spectrum, less the false modal component,and effectively resolve decomposition distortion caused by the endpoint effect.更多还原