改进的LMD方法及其在滚动轴承故障诊断中的应用研究

针对局部均值分解（local mean decomposition, LMD）方法存在的端点效应问题，提出一种基于梯度变化的端点效应抑制方法对局部均值分解进行改进，通过仿真对比不同端点抑制方法的效果，证明了该方法的准确性；针对滚动轴承故障振动信号为一系列调制信号的特点，将改进的局部均值分解方法应用于滚动轴承故障诊断中；利用奇异值分解降噪方法降低噪声污染对分解结果的影响；通过实验验证了该方法在滚动轴承故障诊断中的有效性和可行性。

The improved LMD algorithm and its application in bearing fault diagnosis

To decrease the error induced by the boundary effect in the process of LMD, a new method is introduced which is based on the grades changing. After comparison with other methods by simulation, this method is proved to be more accurate. As the vibration signal of the faulty rolling element bearing is composed by a series of modulating signal, the improved LMD is applied to the fault diagnosis of bearing. This application, in which the SVD is used for noise reduction, is proved to be effective and feasible by experiment.