基于时间序列差分特性的LMD端点效应处理方法

局部均值分解方法在计算过程中会使信号在端部产生端点效应，为减弱LMD中存在的端点效应，提出了一种基于信号时间序列差分特性的端点延拓方法。将信号曲线看作一段离散函数，利用该函数段末端的一阶差分和二阶差分的符号判断曲线的特征和发展趋势，根据具体情况判定所延拓极值点的值。使用该方法对仿真信号和实验轴承外圈故障和内圈故障信号进行了分解，并将其结果与镜像延拓改进的LMD方法分解的结果进行了对比。该方法在分解中的良好表现验证了其在信号分析中的有效性。

A Processing Method to Weaken End Effects of LMD Based on Difference Performance of Time Series

End effects occured at the ends of signals after calculating their envelope and mean value estimation with LMD algorithm. In order to weaken the influences, a continuation approach was proposed based on the difference performance of time series. It considered the signals as a discrete function, and judged the signal tendency by the signs of their first-order difference and second-order difference, determining the value of extreme points of continuation according to the specific situation.This new method was used to decompose the synthetic signals and test signals of rolling bearings with outer raceway faults and inner raceway faults, their results were compared with those decomposed by the LMD improved by mirror extension. Its good performance in decomposition validates the effectiveness of this new approach in signal analysis.